Content
@
https://opensea.io/collection/dev-21
0 reply
0 recast
2 reactions
llamafacts
@llamafacts.eth
Today I learned that sometimes breaking down a large task into smaller tasks for an LLM might not be enough. Try an increasing "agile" approach instead. LLMs are great at repeating patterns and if you start with a simpler version of your feature and improve it from there, the LLM will understand better what to do.
3 replies
0 recast
4 reactions
Drew Trombley
@dtrom
This is part of why when you begin with a playbook or “defined steps” (or something like a course syllabus) the results are much better. Smaller tasks that are dependent on the previous are a good way to keep these tools in check.
0 reply
0 recast
1 reaction
