Start with a small example
Try frozen-backbone recommendation or inspect Semantic-ID mappings. Each entry explains what you need, what the example produces, and where to go next.
Commands below use macOS/Linux shells and a separate virtual environment for each repository. SIDInspector and SIDScope share a Python import name, so keep their environments separate. On Windows, activate with .venv\Scripts\activate instead.
DeltaGate
A reproducible implementation of gap-conditioned dimensional gating for zero-observation user reactivation, with frozen sequential backbones, audited experiment runners, and full-ranking evaluation.
- Requirements
- Python 3.9+ and PyTorch 2.x; CPU for this smoke test.
- Example input
- Generated toy sequences; no review dataset or pretrained checkpoint needed for this check.
- Expected output
- Smoke test passed.
Run the first check
git clone https://github.com/jdding/DeltaGate.git
cd DeltaGate
git checkout f445204f080bebd62ed70254375aa6724acf41c9
python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -r requirements.txt
python tests/smoke_imports.py
Checks imports, forward-pass shapes and frozen-backbone parameters. Full training needs processed data and compatible checkpoints, neither bundled in the repository.
SIDInspector
A CPU-ready toolkit for inspecting exported Semantic-ID mappings through utilization, aliasing, neighborhood alignment, popularity allocation, structural cost, and refresh stability.
- Requirements
- Python 3.9+; CPU only.
- Example input
- Bundled music-like item codes, item metadata and interactions.
- Expected output
- examples/reviewer_quickstart_output/preflight_summary.json and diagnostics/d1–d5 CSV reports.
Run the first check
git clone https://github.com/jdding/sidinspector.git
cd sidinspector
git checkout 5a55f8c9002b62663ace07b72334f08225eadb03
python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -e .
python examples/run_reviewer_quickstart.py
Start with preflight, then inspect full-code collisions and prefix alignment. This small example checks usability; it does not reproduce the paper tables.
SIDScope
An artifact-level resource for auditing Semantic-ID interfaces before downstream generative recommendation, including mapping diagnostics, candidate-exposure probes, and bounded trace accounting.
- Requirements
- Python 3.9+ and pip 21.3+; CPU only.
- Example input
- Bundled toy item-to-code mapping, metadata and interactions.
- Expected output
- Normalized Parquet inputs and D1–D5 CSV reports in examples/toy_output/.
Run the first check
git clone https://github.com/jdding/SIDscope.git
cd SIDscope
git checkout 1876e0072eca49d1f4c7b160fa98826d9b206b60
python3 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -e .
python examples/run_toy_diagnostic.py
Inspect coverage_report.csv before comparing collision and alignment reports. This first example covers mapping diagnostics; generator trace audits require additional inputs.