HuleEdu public research code

This directory defines the generated curated research-code edition of HuleEdu. HuleEdu is the only authored source for this edition.

Approved changes to these selected files are published automatically to the Codeberg repository and the static code reader. The full public commit SHA identifies the same edition in both places.

The edition contains the reviewed research execution sources for feature contracts and extraction, training and score mapping, comparative judgement, Bradley--Terry aggregation, and local embeddings. It contains no research inputs, model packages, provider payloads, or operational configuration.

Run the deterministic synthetic demonstration after installing the declared public dependencies:

python3 -m pip install -r public_research_code/requirements.txt
python3 public_research_code/run_synthetic_demo.py

The command imports the current HuleEdu feature registry, prediction-mapping, and Bradley--Terry modules. It uses only literal synthetic numeric inputs and does not create files, start services, contact a network endpoint, or load a model.

The package is licensed under Apache-2.0; see the root LICENSE file.