4 ms·
But it write mostly useless documentation Which take time to read and decipher. And worse, if you are using it for public documentation, sometimes it hallucina
by orwin 9mo ago
But it write mostly useless documentation Which take time to read and decipher.
And worse, if you are using it for public documentation, sometimes it hallucinate endpoints (i don't want to say too much here, but it happened recently to a quite used B2B SaaS).
- throwup238 9mo agoLoop it. Use another agent (from a different company helps) to review the code and documentation and call out any inconsistencies. I run a bunch of jobs weekly to review docs for inconsistencies and write a plan to fix. It still needs humans in the loop if the agents don’t converge after a few turns, but it’s largely automatic (I baby sat it for a few months validating each change).
- trente 9mo agoThe problem with looping is that any hallucination or incorrect assumption in an early loop becomes an amplifying garbage-in-garbage-out problem.
- tempodox 9mo agoTo translate your answer: - “You’re not spending enough money” - “You’re not micromanaging enough” Seriously?
- joquarky 9mo ago[flagged]
- orwin 9mo agoThat might work for hallucinations, that doesn't work for useless verbose. And the main issue is that LLM don't always distinguish useless verbose from necessary one, so even when I ask it to reduce verbose, it remove everything save a few useful comments/docstring, but some of the comments that were removed I deemed useful. Un the end I have to do the work of cutting verbose manually anyway.
- isjsjsjqhs 9mo ago[dead]
- pocksuppet 9mo agoIt can generate useful documentation or useless documentation. It doesn't take very long to instruct the LLM to generate the documentation, and then check if it matches your understanding of the project later. Most real documentation is about as wrong as LLM-generated documentation anyway. Documenting code is a language-to-language translation task, that LLMs are designed for.
- isjsjsjqhs 9mo ago[dead]