|
|
|
|
|
|
by Sxubas
167 days ago
|
|
|
Just an open thought, what if most improvement we are seeing is not mostly due to LLM improvements but to context management and better prompting? Ofc the reality is a mix of both, but really curious on what contributes more. Probably just using cursor with old models (eww) can yield a quick response. |
|
|