The FAQ says shuffled names (renaming a variable 'count' to 'result') make LLM perform poorly. But I never seen any codebase have this kind of lies (except from comments). And LLM writing code almost never do that.…
The FAQ says shuffled names (renaming a variable 'count' to 'result') make LLM perform poorly. But I never seen any codebase have this kind of lies (except from comments). And LLM writing code almost never do that.…