How namtt908/saint- works
Project type and purpose
• This is a machine-learning library and training pipeline implementing the SAINT+ Transformer-based model for knowledge tracing on the EdNet dataset.
• It is not a web or full-stack app but a PyTorch-based research codebase with data processing, model, trainer, metrics, and CLI/script interfaces.
File structure
• Top-level:
• configs/ednet_config.yaml – hyperparameters and paths
• requirements.txt, setup.py – packaging and dependencies
• saint_plus/ – core library
– data/dataset.py – dataset loader & preprocessing
– model.py – model definition (encoder, decoder, temporal layers)
– trainer.py – training & evaluation orchestration
– utils/metrics.py – performance metrics
• scripts/ – entry-point scripts:
– preprocess_kt1.py – any extra preprocessing
– train.py, train_kt1.py – training launches
• tests/ – unit tests for model behavior
README for extra insights
• README spells out the core components: Encoder, Decoder, Temporal Features, Output Layer, and training/evaluation utilities.
• It cites use of PyTorch 1.10+, details input features and output metrics.