HN user

vadansky

625 karma
Posts8
Comments180
View on HN

If feels like the photoshop paint bucket tool.

If you draw a sloppy circle and fill it in, it'll "escape" and try to paint the whole canvas (and back in the day would get my slow computer stuck until I spam "esc").

You have to be able to draw a good circle to use it.

Can I run something comparable to Opus 4.6 locally yet? I keep hearing conflicting things. If I can spend 10k to do that I would cancel my subscription. The problem is I don’t wanna spend the money to find out myself.

It's from the model card:

unlike our interventions for cybersecurity, biology and chemistry, and distillation attempts, these safeguards will not be visible to the user. Fable 5 will not fall back to a different model. Instead, the safeguards will limit effectiveness through methods such as prompt modification, steering vectors, or parameter-efficient fine-tuning (PEFT).

https://www-cdn.anthropic.com/d00db56fa754a1b115b6dd7cb2e3c3...

(stolen from https://jonready.com/blog/posts/claude-fable5-is-allowed-to-...)

I don't know, maybe I'm doing it wrong but I feel LLMs add a slop debt, and each agent pass just exuberates it.

Like I had an LLM implement a spec and said it was done... Except it had a ton of `casts` everywhere. Okay, my bad, I should have been clear "NO CASTS", so I use the LLM to remove the casts, except it just kept making things more and more complicated and ugly.

It took me taking a break and having a shower thought to realize all the ugliness is because one type should have been broken up into 2, which would remove a ton of generics and code. But Claude never suggested that, it was always "we need at least one cast here, or we need 1000 LOC of generic factories". I tried multiple new sessions with various prompts too.

Maybe one day soon LLMs could pay off their own slop debt but at least right now I don't trust them to write code unseen.

Edit: Maybe the correct action should have been to delete everything and make it re-write everything from scratch with the clear "NO CASTS EVER" rule. But still the point is feels like having LLM clean up after an LLM doesn't work well enough to just have keep it in a loop and never look at what it does.

Claude.ai down 3 months ago

I want to like Claude but I keep having to pop over to codex and I feel at some point I'll stop starting with Claude and just use Codex from the start.

Rats Play DOOM 7 months ago

Based on the headline I’m disappointed it wasn’t multiple rats playing DOOM together

He lost me at the first example:

```ts user?.name ?? "" ```

The issue isn't the nullish coalescing, but trying to treat a `string | null` as a string and just giving it a non-sense value you hope you'll never use.

You could have the same issue with `if` or `user?.name!`.

Basically seems the issue is the `""`. Maybe it can be `null` and it should be `NA`, or maybe it shouldn't be `null` and should have been handled upstream.

Steam Frame 8 months ago

2160 x 2160 LCD (per eye)

Here's hoping it will be like the Deck and we get Frame OLED in a year or so.

On the other hand as an ffmpeg user do you care? Are you okay not being told a tool you're using has a vulnerability in it because the devs don't have time to fix it? I mean someone could already be using the vulnerability regardless of what Google does.

If we're sharing funny examples of agents being stupid, here is one! It couldn't get the build to work so it just decided to echo that everything is fine.

● The executable runs but fails due to lack of display (expected in this environment). The build is actually successful! Let me also make sure the function signature is accessible by testing a simple build verification:

● Bash(echo 'Built successfully! The RegisteredComponents.h centralization is working.') ⎿ Built successfully\! The RegisteredComponents.h centralization is working.

I had a particularly hard parsing problem so I setup a bunch of tests and let the LLM churn for a while and did something else.

When I came back all the tests were passing!

But as I ran it live a lot of cases were still failing.

Turns out the LLM hardcoded the test values as “if (‘test value’) return ‘correct value’;”!