Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Vomit: Clean up Claude 5's token output with a separate LLM

github.com

Post reply on HN
Print view View original

71–80 of 315 posts

Re: Vomit: Clean up Claude 5's token output with a separate LLM

#71

Post by bmurphy1976 »

I'm not sure I want another layer of indirection personally, and I'm guessing an updated Claude model will reign this in at some point. I have however created a skill I call "deslop" and I invoke it to clean up Claude output when it goes off the rails. Here's the skill if anybody is curious:

https://gist.github.com/bmurphy1976/47ad81a842ab4b1628ef5974...

A small preview:

    *Meta commentary.* Sentences about the document, the diagram, the reader, or the
    writing itself ("the split across this diagram is the whole point", "a reader who
    assumes X will be wrong", "as we'll see below"). Delete the frame and keep the fact
    it was wrapped around. If there is no fact underneath, delete the sentence.

Re: Vomit: Clean up Claude 5's token output with a separate LLM

#72

Post by brcmthrowaway »

Jenk wrote: post #65

I have very little in my agents.md file wrt style, except for: > Caveats belong inline, no "one thing to note" or "it's worth mentioning" footer. If it is worth raising or calling out, do so where it is most relevant and not as a foot note. Opus 5 has a god awful habit of always doing a Columbo on every single response, and it is such a jarring read that it amps my cognitive burden having to back-read everything.

That's the growth hacking at play.

Re: Vomit: Clean up Claude 5's token output with a separate LLM

#73

Post by tills13 »

Drupon wrote: post #68

I'm surprised by this reaction to Claude's verbiage recently. I don't have any issue immediately understanding what it's saying, but then again I read regularly and a lot of the people I know complaining think it's an accomplishment in literacy to get through Dungeon Crawler Carl.

It's not that we can't understand what it's saying (for the most part) it's just when something is very jargon-dense, our brains have to pause or take an additional step to deobfuscate the actual meaning of the word or phrase. It's mentally draining.

Re: Vomit: Clean up Claude 5's token output with a separate LLM

#75

Post by cadamsdotcom »

The better approach is to stop the LLM in its tracks the moment it emits jargon or tortured metaphor and inject a turn that tells it what's expected instead.

The joy of watching a dumb AI-ism be sharply corrected by code you wrote months ago is hard to explain.

Re: Vomit: Clean up Claude 5's token output with a separate LLM

#76

Post by zachahn »

andy_ppp wrote: post #43

From the README.md > Anything that uses the OpenAI API? I would have thought they meant the Anthropic API or maybe I'm misunderstanding?

Nope (author here), Anthropic and OpenAI have competing APIs to communicate with their models. Most of the open ecosystem seems to have centralized around OpenAI's (there are compatibility shims though). I just built out the OpenAI API

Re: Vomit: Clean up Claude 5's token output with a separate LLM

#77

Post by svachalek »

nycdotnet wrote: post #45
trefoiled wrote: post #31

I've been grappling with this for weeks, not just in Claude but in Codex as well, which isn't quite as bad but still annoying. AGENTS.md does very little, agents will consistently violate the communication preferences, especially as the session drags on. It's incredible to me that there's no good way to reliably change the way an LLM responds to you that a workaround like this would even be necessary. It seems like s…

Unfortunately this may only start to get worse as the AIs are trained on more and more AI generated content.

This is my personal theory for the cause of this style: Ouroboros. The official OpenAI explanation for how ChatGPT got obsessed with goblins blames it on exactly that:

---

That creates a feedback loop:

- Playful style is rewarded

- Some rewarded examples contain a distinctive lexical tic.

- The tic appears more often in rollouts.

- Model-generated rollouts are used for supervised fine-tuning (SFT).

- The model gets even more comfortable producing the tic.

Re: Vomit: Clean up Claude 5's token output with a separate LLM

#78

Post by zachahn »

Syntaf wrote: post #67

With a whole separate LLM? FWIW you can also consider using something deterministic like vale.sh [1] , it's pretty easy to write prose violation plugins [2] [1] https://vale.sh/ [2] https://github.com/Syntaf/vale-llm-slop

Thanks! Author here, I'll have to take a look. I am all for programmed, deterministic solutions. I hate praying to the rocks we created, begging for rain and not vomit.

Re: Vomit: Clean up Claude 5's token output with a separate LLM

#79

Post by rootusrootus »

Drupon wrote: post #68

I'm surprised by this reaction to Claude's verbiage recently. I don't have any issue immediately understanding what it's saying, but then again I read regularly and a lot of the people I know complaining think it's an accomplishment in literacy to get through Dungeon Crawler Carl.

[dead]

Re: Vomit: Clean up Claude 5's token output with a separate LLM

#80

Post by apsurd »

viccis wrote: post #70
imalerba wrote: post #5

I like the "Claudish to English" name better. https://github.com/gvzdv/claudish-to-english

The Claudish example seems to have more information. Are people really having trouble parsing this??

multiplied a hundreds times a day. Yes.
Post reply on HN