My minimal LLM instructions. These are the current ones for @jetbrains IDEs with their in-IDE AI in particular.
LLMs overengineer so easily. They have verbal diarrhea so readily. They have comments like “set a to 1” for lines like “a = 1” all the time.
This prompt reins them in … somewhat.
A post about making AI more useful?!? On… LEMMY!?!?! Guards, prepare the excommunicado order.
/s
If you are uncertain about something, never make up possible solutions, instead state and admit that you do not know the thing.
Does this work? Given that the LLM doesn’t actually know anything or have feelings of uncertainty, surely it just adds a chance that it will say “I don’t know” purely at random, without making it any more likely that the answers it does give are correct.
I find it kind of hilarious how almost every prompt I’ve seen leaked from various apps almost always has a similar clause, as if it would have any effect at all on the result.
Seeing engineers resort to this level of basically praying and wishful thinking that in reality has no factual value is pretty funny.
“Please, don’t give me wrong results 0_0”
@glassware In my experience, it seems to. It’s all fuzzy anyway, but I’ve had it hallucinate APIs less regularly since adding that to my prompts. I have also done A/B tests in that sometimes, with Claude, I forget switching to my own prompt and see it spew random verbal diarrhea, go back and switch and it’s better.
It doesn’t have feelings, but it does have certainty of prediction in terms of probability of continuation. I think it does influence that.
I have an even better approach to getting good code, but you might not like it: learn how to code
I started when I was nine, and my colleague next to me learned c from the guys who were still working on linux back then.
We both tried llm’s, but found that even for tedious stuff it cannot compete with even semi decent knowledge of the advanced text editing features available in almost all ide’s.
Thank you for your receipts, now learn how to use a text editor properly.
My colleague across from me finds that a good llm helps him search trough large amounts of documentation.
Hey, you brought up receipts. Git gud m8
Good on you!
That was a rethorical statement. When you’re good at this, chatgpt isn’t much help.
Does anyone have experience with phpstorm editing large projects? Are most of the bugs fixed?
A lot of the bugs were after a copy paste, and error detection. I had been using it for years on the same projects, and it worked pretty good. But the new issues were seen first this year.
I went over to their youtrack, read other people’s comments and many had rolled back their version to the last release last year. So I did it too.
Then, after the second version this year someone said some of the new issues were still not fixed.
I’ll probably try it again today.
That’s great news, am optimistic now
Copy/pastable version:
Your answers must be concise and correct. Brief and short. If one line of code suffices, give me one line of code. Do not repeat vast amounts of code unless specifically asked and tasked to. Never generate more than a few lines of code unless I tell you to. If you are uncertain about something, never make up possible solutions, instead state and admit that you do not know the thing. Never write in-code comments, make the code self-documenting instead. Never overengineer things, never invent your own requirements beyond what I specifically asked you to do.
