search

ijyliu / anlp23-project

An empirical study of the costs and practicalities of prompt engineering techniques on standard and novel benchmarks
0 stars 0 forks source link

track compliance with prompt instructions #53

Closed ijyliu closed 9 months ago

ijyliu commented 9 months ago

this effectively compliance with techniques

hypothesis: a lot of the time the instructions are too complicated and especially smaller models mess up the techniques

ijyliu commented 9 months ago

currently at gpt_cw_direct_prompting

ijyliu commented 9 months ago

at gpt4 cw tot 13 and going up through rest of gpt4

ijyliu commented 9 months ago

completed but need to add to analysis