[marceloabsousa]

I wouldn't discard the capabilities of the LLM. I've been using it to do relational reasoning of Java comparators and it is surprisingly accurate. https://twitter.com/marceloabsousa/status/163289383288582963...