Hi, I am using LangChain to call Gemini AI models (Gemini 2 Flash, Gemini 2 Pro-Exp, Gemini 1.5 Pro, etc.) The task is to analyze 2500 enumerated ideas and put them in different buckets, according to some rules. One of the rules is that only 25% of the most important ones can go into bucket 1. So I cannot divide the input into smaller files.
My question is regarding the best way for inputing these ideas into Gemini models. Should I use enumerated plain text as input? Does JSON formatted input help? Thanks.