Content
@
0 reply
0 recast
0 reaction
Shashank
@0xshash
lots of great practical insights on llms - smaller prompts work better. gpt actually gets confused if you over-specify. quality can improve when you’re more vague - langchain is pre-mature abstraction - vectordbs are rarely useful outside of search https://kenkantzer.com/lessons-after-a-half-billion-gpt-tokens/
2 replies
0 recast
6 reactions
Sid
@sidshekhar
very interesting. Esp the improved results with more vague prompting claim
0 reply
0 recast
0 reaction
