--- title: Tutor Evaluation Annotation App sdk: docker app_port: 7860 pinned: false license: gpl-3.0 --- # Potato annotation app (math) Annotation platform for the math study (MathVision + OrcaMath), adapted from the programming study app. Copy this folder into a Hugging Face Docker Space. The main files are: - `hf_start.py`: participant login, subset assignment, progress recovery, quality checks, result backups, Prolific redirects, the `/mathvision_images/` route, and the final-item save fix; - `config.yaml`: task text, annotation dimensions, guidelines/video links, interface styling, and KaTeX math rendering; - `my-annotation-task/data/`: the Potato-ready CSV and subset manifest; - `my-annotation-task/images/mathvision/`: MathVision problem images (tracked with Git LFS, see `.gitattributes`); - `tools/fetch_mathvision_images.py`: downloads the MathVision images; - `tools/create_prolific_potato_ready_data.py`: prepares all data files; - `Dockerfile` and `requirements.txt`: the Hugging Face setup. Secrets: `HF_TOKEN`, `PROLIFIC_SUCCESS_COMPLETION_URL`, `PROLIFIC_FAILED_TEST_COMPLETION_URL`. Upload `../data/gold_annotations.json` to the private results repository configured in `config.yaml`. Study design, rebuild steps, and the pre-launch checklist are in [`../README.md`](../README.md).