Suite20 package =============== check_suite20_environment.py Validates Python/PyTorch and canonical + structured cache alignment before training. suite20_data.py Shared cross-faculty loader, chronology split, scaling and imbalance sampler utilities. suite20_models.py Flat control, GRU temporal hybrid, and primary target-conditioned Semester Transformer. suite20_training.py Weighted shared-faculty training, chronological epoch selection, production refit and inference. suite20_reporting.py Metrics, old-suite/Council/V2 comparison, temporal uplift, attention/gate/uncertainty diagnostics. train_suite20_semester_target_transformer.py Main experiment entry point. run_suite20.sh Production server run command. README.md Run instructions and result interpretation. DESIGN.md Frozen scientific design and acceptance criteria. requirements.txt Minimal runtime dependencies. tests/ Core model/data/reporting tests.