More and more I find code like this. When you dig into an interesting aspect it's just an llm call. I haven't decided yet if it's good or bad. But it does seem to work a most of the time.
The problem is that we're used to code that works all the time or none of the time. With LLMs you can't know if it's right for sure and it's really hard to check.
Clarifying that it’s the output of an LLM is responsible.
Though LLMs are really good at anything related to plain LeetCode problems. There has been so much written about the standard LeetCode problems across so many websites that it’s all heavily represented in training sets.
Yeah sorry :) At first I tried to do a stress test, then plot a curve and try to find the closest runtime curve to that curve but it didn't work most of the time ;-;
The problem is that we're used to code that works all the time or none of the time. With LLMs you can't know if it's right for sure and it's really hard to check.