Report
An 8K context budget reportedly hit 98.25% F1 in financial-document classification
A user says the 8K budget had 84% lower model cost than their 480K reference using OpenAI's Decisions API.
TLDR
A user says an 8K context budget was enough to classify financial documents at 98.25% F1, with 84% lower model cost than their 480K reference. They point to the result as an example of not needing to use all the context available through OpenAI's Decisions API.
Combined views
—
2 Sources, first seen ago
