How to reproduce these experiments
Everything on this page is captured automatically — nothing here is a manually typed claim of reproducibility.
Recorded environment
| Item | Recorded value |
|---|---|
| Random seed | 42 |
| Python version | 3.11.9 (tags/v3.11.9:de54cf5, Apr 2 2024, 10:12:12) [MSC v.1938 64 bit (AMD64)] |
| OS / runtime | Windows-10-10.0.29617-SP0 |
| Git commit (environment capture) | 4130c370545f |
| Git commit (this evidence export) | d4c1b0b37dd4 |
| Key package versions | numpy 2.4.6, pandas 3.0.5, scipy 1.17.1, scikit-learn 1.9.0, matplotlib 3.11.1, PyYAML 6.0.3 |
Quick start
python -m venv .venv
python -m pip install -r requirements.txt
python scripts/download_dataset.py
python scripts/validate_dataset.py
python scripts/run_drift_analysis.py
pytest -qFull command reference: docs/REPRODUCIBILITY.md · docs/RESEARCH_PROTOCOL.md
Evidence policy
Every number on this site traces to a saved artifact and a specific experiment ID. If an artifact does not exist, the corresponding value renders as Not executed — never a placeholder or estimated figure. Synthetic data is limited to unit tests and is never shown as a research result.
Google Colab
Real Colab CLI execution is orchestrated from WSL2 via scripts/colab/ and VS Code tasks (Mode B), or interactively via Select Kernel → Colab (Mode A). Current verified status: results/reproducibility/COLAB_STATUS.md.
Cross-platform provenance bridge
GitHub, Hugging Face, and Kaggle CLI status, captured live (never fabricated):
| Platform | Authenticated | CLI version |
|---|---|---|
| GitHub | Yes | gh version 2.97.0 (2026-07-31) |
| Hugging Face | No | version=1.27.0 |
| Kaggle | No | Kaggle CLI 2.2.4 |
Full policy: docs/RESEARCH_PLATFORM_BRIDGE.md