These are some words I’m writing to test the content width of this webpage. I hope the width I have set isn’t so large that it becomes uncomfortable to read, but on the other hand I hope it isn’t too small either.
This tree shows one description broken down to its smallest constituent parts. It visualizes a task that Gemini Advanced was unable, or perhaps too lazy, to do itself.
Extraction: the prompt first asks the models to identify those entities in the text that meet the criteria outlined in the prompt. In this case, that includes the names individual items which Tom Sawyer received as payment for the privilege of letting the other children add a layer of white paint to Mrs. white picket fence, as well as the quantity of those items.
Classification: the prompt then asks the models to classify each obkject “accordin to how each is used.” I did not specify the categories I wanted the models to use to categorize these object, because for this test I wanted to see what the models came up with on their own. [difference between structured/unstructured categories?]