[dead]
Universal Claude.md – cut Claude output tokens
51–60 of 192 posts
Re: Universal Claude.md – cut Claude output tokens
#52I might be wrong but based on the videos I've watched from Karpathy, this would, generally, make the model worse. I'm thinking of the math examples (why can't chatGPT do math?) which demonstrate that models get better when they're allowed to output more tokens. So be aware I guess.
I think that concern is valid in general terms, but it’s not clear to me that it applies here. The goal here seems to be removing low-value output; e.g., sycophancy, prompt restatement, formatting noise, etc., which is different than suppressing useful reasoning. In that case shorter outputs do not necessarily mean worse answers. That said, if you try to get the model to provide an answer before providing any reasoni…
The file starts with:
> Answer is always line 1. Reasoning comes after, never before.
> No explaining what you are about to do. Just do it.
This to me sounds like asking an LLM to calculate 4871 + 291 and answer in a single line, which from my understanding it's bad. But I haven't tested his prompt so it might work. That's why I said be aware of this behavior.
Re: Universal Claude.md – cut Claude output tokens
#53Lol this is so naive and optimistic. Claude will just do whatever it wants and apologize later. This is good for action #1 though.
Re: Universal Claude.md – cut Claude output tokens
#54Things like this make me sad because they make obvious that most people don’t understand a bit about how LLM work. The “answer before reasoning” is a good evidence for it. It misses the most fundamental concept of tranaformers: the are autoregressive. Also, the reinforcement learning is what make the model behave like what you are trying to avoid. So the model output is actually what performs best in the kind of soft…
To me it’s as simple as “who knows best how to harness the premier LLM – Anthropic, the lab that created it, or this random person?”
That’s why I’m only interested in first party tools over things like OpenCode right now.
Re: Universal Claude.md – cut Claude output tokens
#55As with all of these cure-alls, I'm wary. Mostly I'm wary because I anticipate the developer will lose interest in very little time and also because it will just get subsumed into CC at some point if it actually works. It might take longer but changing my workflow every few days for the new thing that's going to reduce MCP usage, replace it, compress it, etc is way too disruptive. I'm generally happy with the base Cl…
The hidden cost with all of these "fix Claude" layers is that your workflow keeps moving underneath you.
Even when one helps, you're still betting it won't be obsolete or rolled into the defaults a few weeks from now.
Re: Universal Claude.md – cut Claude output tokens
#56Paul Kinlan published a blog post a couple of days ago [1] with some interesting data, that show output tokens only account for 4% of token usage. It's a pretty wide-reaching article, so here's the relevant quote (emphasis mine): > Real-world data from OpenRouter’s programming category shows 93.4% input tokens, 2.5% reasoning tokens, and just 4.0% output tokens . It’s almost entirely input. [1]: https://aifoc.us/the-…
However output tokens are 5-10 times more expensive. So it ends up a lot more even on price
Even more than that in practice once you factor in prompt caching
Re: Universal Claude.md – cut Claude output tokens
#57> Answer is always line 1. Reasoning comes after, never before.
The very first rule doesn’t work. If you ask for the answer up front, it will make something up and then justify it. If you ask for reasoning first, it will brainstorm and then come up with a reasonable answer that integrates its thinking.
Re: Universal Claude.md – cut Claude output tokens
#58So there's a direct monetary cost to this extra verbiage:
"Great question! I can see you're working with a loop. Let me take a look at that. That's a thoughtful piece of code! However,"
And they are charging for every word! However there's also another cost, the congnitive load. I have to read through the above before I actually get to the information I was asking for. Sure many people appreciate the sycophancy it makes us all feel good. But for me sycophantic responses reduce the credibility of the answers. It feels like Claude just wants me to feel good, whether I or it is right or wrong.
Re: Universal Claude.md – cut Claude output tokens
#59> No unsolicited suggestions. Do exactly what was asked, nothing more.
> No safety disclaimers unless there is a genuine life-safety or legal risk.
> No "Note that...", "Keep in mind that...", "It's worth mentioning..." soft warnings.
> Do not create new files unless strictly necessary.
Nah bruh. Those are some terrible rules. You don't want to be doing that.
Re: Universal Claude.md – cut Claude output tokens
#60[dead]