[–] jneagu 1y ago ↗ Nice! Overall, frameworks that re-prompt an LLM with feedback or failure modes of the original outputs do very well.
1 comment
[ 4.2 ms ] story [ 13.7 ms ] thread