Apppublic

sanibusiness1236/integrity-ml

sourceHugging Facemitupdated 4mo agoView on Hugging Face

0likes

App README

INTEGRITY ML Service

GNN-based academic integrity prediction service used by the INTEGRITY platform.

What this Space exposes

Endpoint	Description
`GET /health`	Service health and current mode
`GET /models`	List available GNN architectures and which one is active
`POST /predict`	Run an integrity prediction over a venue graph
`POST /evaluate`	Run evaluation metrics over a labelled venue graph
`GET /docs`	Auto-generated Swagger UI

Available models

The registry pre-loads four graph neural network architectures with checkpoints committed to this repo so cold starts are fast:

—Vanilla GCN (default)
—H2GCN — heterophily-aware
—FAGCN — frequency-adaptive
—GraphSAGE

Local development

bash

python -m venv .venv
.\.venv\Scripts\activate          # (Linux/macOS: source .venv/bin/activate)
pip install -r requirements.txt
uvicorn app.main:app --reload --host 0.0.0.0 --port 8000

Deployment

This folder is shaped for Hugging Face Spaces (Docker SDK). The Dockerfile at the root of this directory is what HF will build. The trained .pt files under trained_models/ are committed so the service is ready to serve predictions the moment the container starts.