This software tests an iterative propose, validate and select loop that improves keyword rules for text classification on a small synthetic dataset.
-
Updated
Oct 5, 2026 - Python
This software tests an iterative propose, validate and select loop that improves keyword rules for text classification on a small synthetic dataset.
This project trains and compares computer models that reuse layers to classify whether an on/off sequence contains an odd or even number of on values.
To associate your repository with the iterative-adaptation topic, visit your repo's landing page and select "manage topics."