Core One Troubleshooting Guide
For fun I asked Claude to parse all 20K posts here and create a Core One Trouble Shooting Guide. So here it is (see attached). I know how some of you feel about AI but have a look, I think it did a pretty good job. I certainly don't have the time nor patience to do this myself. Claude took 2m37s to produce this document reading the entire forum.
RE: Core One Troubleshooting Guide
Version 2 which includes the Prusa Knowledge Base. I'm going to carefully read this document tonight and manually clean up any AI slop
RE:
"Visible" fine artifacts 🙂
"print at higher speeds" seems misleading to me.
RE: Core One Troubleshooting Guide
"Visible" fine artifacts 🙂
"print at higher speeds" seems misleading to me.
Agree. I'm going to read through it, take notes and make corrections.
RE: Core One Troubleshooting Guide
Don't get me wrong, it's a great job, thanks for the effort. But frankly, there is less understanding than I'd have thought e.g. "overconstraint" in Z-axis context, no comprehension at all.
But it may be simply too little computing time for too much output - when I run AI, it grinds wheels for 30 seconds on a single detail question.
RE:
Don't get me wrong, it's a great job, thanks for the effort. But frankly, there is less understanding than I'd have thought e.g. "overconstraint" in Z-axis context, no comprehension at all.
But it may be simply too little computing time for too much output - when I run AI, it grinds wheels for 30 seconds on a single detail question.
Agree. I'm going to manually edit and strip out anything that is nonsensical. To be honest, I haven't put much effort into this (yet) at all other than telling it what to do. I asked it not to make any inferences beyond what is here, in the prusa knowledge base and on reddit (this last one is probably a mistake). What I might do is convert this to a public wiki and we can take turns updating it.
I'm thinking of a Wikipedia style home for it.
RE: Core One Troubleshooting Guide
Don't get me wrong, it's a great job, thanks for the effort. But frankly, there is less understanding than I'd have thought e.g. "overconstraint" in Z-axis context, no comprehension at all.
But it may be simply too little computing time for too much output - when I run AI, it grinds wheels for 30 seconds on a single detail question.
... I'm going to manually edit and strip out anything that is nonsensical.
Yes but is this still going to be fun? Launching an AI summary query is one thing, good idea, interesting experiment.
Now managing a book "what you always wanted to know about <enter printer name here> but never dared to ask" even as editor (for an imbecile author) sounds like a job people usually get paid for.
RE:
Yes but is this still going to be fun? Launching an AI summary query is one thing, good idea, interesting experiment. Now managing a book "what you always wanted to know about but never dared to ask" even as editor (for an imbecile author) sounds like a job people usually get paid for.
I'm thinking of hosting it here: https://miraheze.org/ as a Wiki page and we can all contribute. Or otherwise maybe Github wiki is better? I'll rapidly lose interest if I maintain it myself.
RE: Core One Troubleshooting Guide
Next version. I'll stop here and move it to a wiki and add some human touches to it
RE: Core One Troubleshooting Guide
just make sure some poor innocent AI, new to this world with eyes full of wonder doesn't accidentally choke on AI-generated content 😀
RE: Core One Troubleshooting Guide
just make sure some poor innocent AI, new to this world with eyes full of wonder doesn't accidentally choke on AI-generated content 😀
I've done some crazy stuff recently like asking ChatGPT to create a script for me to give to Claude to create an application. ChatGPT is terrible at coding but is better than Claude at writing a spec. Anyway, these tools are facinating. It would have taken me multple days/weeks to put this document together. Actually I would have never put something like this together. I have some time yet before I retire and have free time at my disposal.
Even if it's 20-30% wrong or misleading it's still not bad.
RE: Core One Troubleshooting Guide
[...] I know how some of you feel about AI but have a look, I think it did a pretty good job. [...]
m2c: Thx for initiating such AI job which will help us. This may solve one of the problems when looking/searching for a solution for a specific problem: You are standing in front of a jungle of docs/posts etc. and do not know how to find them - "Which specific search phrase?", "Where? Help Docs, Forum...?"
Now, it's time to think about how to implement such "AI Summary function" ...
Thx again!
Core One L
RE: Core One Troubleshooting Guide
This may solve one of the problems when looking/searching for a solution for a specific problem: You are standing in front of a jungle of docs/posts etc. and do not know how to find them - "Which specific search phrase?", "Where? Help Docs, Forum...?"
You can always ask the language model of your choice, telling it to use the Prusa forum as its main information source. The LLMs are quite good at making sense of questions where you describe the issue and don't know the "correct" search term. They will provide you with a summary of the answer as well as links to the actual sources if you want to dive deeper (or don't fully trust the AI summary).
When I am looking for an answer to a specific problem, I prefer this targeted approach. But of course there are good uses for a comprehensive AI summary of all Core One troubleshooting discussions as well, e.g. to get an overview and awareness of potential future issues.
RE: Core One Troubleshooting Guide
Thank you for sharing👍
Could you post the prompt for this query ?