Instructions to use SZLHOLDINGS/szl-invariants with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Kernels
How to use SZLHOLDINGS/szl-invariants with Kernels:
# !pip install kernels from kernels import get_kernel kernel = get_kernel("SZLHOLDINGS/szl-invariants") - Notebooks
- Google Colab
- Kaggle
szl-invariants
Eight falsifiable receipt/ledger invariants, offline. Not a model. No weights.
IS: a pure-Python, stdlib-only governance kernel that replays the same eight checks the a11oy backbone runs at /api/invariants over a ledger JSONL you already hold. Public API: run_invariants, load_jsonl, verify_ed25519, selfcheck. Statuses HOLDS / VIOLATED / KEY_ROTATED / NO_DATA / UNAVAILABLE are first-class.
IS NOT: trained weights or a LoRA. Not a CUDA bench. Passing selfcheck is not an eval leaderboard. Hub model.joblib is QUARANTINED executable serialization — do not joblib.load it. These checks do not prove the export is complete.
Canonical GitHub source: https://github.com/szl-holdings/szl-invariants
Hub package: https://huggingface.co/kernels/SZLHOLDINGS/szl-invariants
This model-type repo is the publish / card mirror. Apache-2.0.
from kernels import get_kernel
inv = get_kernel("SZLHOLDINGS/szl-invariants", revision="main", trust_remote_code=True)
report = inv.run_invariants(inv.load_jsonl("runs_export.jsonl"),
samples=inv.load_jsonl("training_samples.jsonl"), pubkey="<SPKI base64>")
print(report["summary"])
print(inv.selfcheck())
The eight ids: receipt-chain-continuity, ledger-failure-shape, served-run-has-model, signed-columns-atomic, loop-steps-positive, receipt-ed25519-verify, receipt-columns-consistent, flywheel-lineage. Missing pubkey → #6 UNAVAILABLE; missing samples → #8 UNAVAILABLE.
Λ is not touched here and stays Conjecture 1. Doctrine v11. Copyright 2026 SZL Holdings · Stephen P. Lutar · ORCID 0009-0001-0110-4173.
- Downloads last month
- -