Thread

Tweet 1

@DeathOrCaffeine @eigen_spectra @_wilfredh This sort of gives a taste of how I use LLMs: I generally try to remember the answers if it feels like basic knowledge. I keep all my interactions in a single text file so it's easily searchable if I want to revisit something

---

Tweet 2

@DeathOrCaffeine @eigen_spectra @_wilfredh If you expect perfect output 100% of the time obviously you'll be disappointed. But I think for personal use code it's first attempt is often fine. And often you can get exactly what you want in one or 2 iterations

---