Comment by nuancebydefault
1 year ago
My takeaway and also personal experience is that you get the best results is that you co-develop with the LLM.
- write a simple prompt that explains in detail the wanted outcome.
- look at the result, run it and ask it how it can improve.
- tell it what to improve
- tell it to make a benchmark and unit test
- run it each time and see what is wrong or can be improved.
One approach I've been using recently with good results is something along the lines "I want to do X, is there any special consideration I should be aware while working in this domain?". This helps me a lot when I'm asking about a subject I don't really understand. Another way to ask this is "What are the main pitfalls with this approach?".
I'm using o1, so I don't know how well it translate to other models.
Same experience.
Also: If you're experienced at code reviews, you can get great results.