In analyze_content.py, data is chunked to avoid exceeding the input limit. This means there are multiple calls to Gemini, where the outputs are appended together. We want a single output object with info for the topics we want
We need to find a prompt which would output a consistent format. (look into prompt engineering for Gemini)
We need to combine the outputs for each chunk into one object
Instead of
{
"Relevant people": Bob
},
{
"Relevant people": Jane
}
Michaelyya
commented
3 months ago
analyze_content.py, data is chunked to avoid exceeding the input limit. This means there are multiple calls to Gemini, where the outputs are appended together. We want a single output object with info for the topics we wantInstead of { "Relevant people": Bob }, { "Relevant people": Jane }
we need
{ "Relevant people": Bob, Jane }