tiny hamlet
Hey folks! I'm working on a project where I want to use a whole bunch of API calls to classify small blocks of text according to a defined JSON schema of Main Category > Sub Category > SubSub Category. (E.g. listing the topics of several thousand tweets).
I want to be able to run the code on multiple different collections of text, each with completely different bunches of categories.
Right now i've been sending the entire JSON structure with each API call, but sometimes that includes hundreds of subsubcategories leading to tons of repeat tokens being used.
I want to be able to run each request in parallel so that the whole thing doesn't take hours each time, which so far I've been able to do.
Is there any way of giving an initial prompt or system instruction and then splitting off into a bunch of parallel requests based on it, each only containing the text I want analyzed?
Thanks!
livid moth