Feature Extraction
Kernels
Scikit-learn
kernel
governance
provenance
suite
embeddings
word-embeddings
doi:10.5281/zenodo.19944926
Instructions to use SZLHOLDINGS/szl-kernels with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Kernels
How to use SZLHOLDINGS/szl-kernels with Kernels:
# !pip install kernels from kernels import get_kernel kernel = get_kernel("SZLHOLDINGS/szl-kernels") - Scikit-learn
How to use SZLHOLDINGS/szl-kernels with Scikit-learn:
from huggingface_hub import hf_hub_download import joblib model = joblib.load( hf_hub_download("SZLHOLDINGS/szl-kernels", "sklearn_model.joblib") ) # only load pickle files from sources you trust # read more about it here https://skops.readthedocs.io/en/stable/persistence.html - Notebooks
- Google Colab
- Kaggle
| { | |
| "schema": "szl.model-source-attestation/v2", | |
| "observed_at": "2026-07-30T19:33:24Z", | |
| "model": { | |
| "id": "SZLHOLDINGS/szl-kernels", | |
| "repository_type": "model", | |
| "artifact_kind": "governed kernel code and configuration with trained SZL-MiniEmbed embeddings", | |
| "trained_weights_present": true, | |
| "surrogate": { | |
| "name": "SZL-MiniEmbed v1", | |
| "files": { | |
| "vectors.npz": "053d839490ea95ed03dd1059f68474ef77cfcf04b2e727b29da9bc617abe4020", | |
| "vocab.json": "7862ebac9193d010606d89fb5aa6c046cbd638590d7a40c068c7bb60b3ed665e", | |
| "config": "config.json" | |
| }, | |
| "role": "intrinsic word-embedding companion over the SZL text estate; the kernel suite remains the primary artifact", | |
| "method": "distance-weighted term-term co-occurrence, PPMI, and TruncatedSVD", | |
| "fidelity_measured": { | |
| "vocab_size": 3290, | |
| "embedding_dim": 128, | |
| "svd_explained_variance_ratio": 0.3146, | |
| "probe_terms_in_vocab": 20, | |
| "claim_scope": "INTRINSIC_SANITY_ONLY" | |
| }, | |
| "receipt": "TRAINING_RECEIPT.json", | |
| "reverify": "python scripts/eval.py" | |
| } | |
| }, | |
| "source_of_record": { | |
| "state": "GITHUB_CANONICAL_SOURCE", | |
| "platform": "github", | |
| "repository": "szl-holdings/szl-kernels", | |
| "url": "https://github.com/szl-holdings/szl-kernels", | |
| "exact_revision_evidence": "publication.json in the Hugging Face artifact", | |
| "publication_contract": "publishing/source-binding.json", | |
| "publication_workflow": "https://github.com/szl-holdings/szl-forge/blob/main/.github/workflows/publish-szl-kernels.yml" | |
| }, | |
| "artifact_history": { | |
| "imported_from": "https://huggingface.co/SZLHOLDINGS/szl-kernels", | |
| "imported_hub_revision": "46730d363d1b2cf69c1ef770c60bb84a3356b574", | |
| "history_preserved": true | |
| }, | |
| "claims": { | |
| "github_source": "CANONICAL", | |
| "huggingface_source_binding": "EXACT_GIT_REVISION_AFTER_MAIN_PUBLICATION_AND_READBACK", | |
| "artifact_equivalence": "BYTE_IDENTICAL_DECLARED_FILE_SET_AFTER_PUBLICATION", | |
| "reproducible_build": "BOUNDED_REPLAY_ONLY", | |
| "trained_model": "TRAINED_EMBEDDINGS_PRESENT", | |
| "independent_quality_certification": "NOT_CLAIMED", | |
| "serving_process_revision": "NOT_APPLICABLE_REPOSITORY_ARTIFACT" | |
| }, | |
| "replay_contract": { | |
| "dependencies": "requirements-replay.txt", | |
| "checks": [ | |
| "shipped vectors and vocabulary SHA-256 match TRAINING_RECEIPT.json", | |
| "mean top-k neighbor-set Jaccard overlap is at least 0.90", | |
| "SVD explained-variance delta is at most 0.02" | |
| ], | |
| "byte_identical_retraining": "NOT_CLAIMED" | |
| }, | |
| "limits": [ | |
| "SZL-MiniEmbed is a NumPy word-embedding table, not a transformer language model.", | |
| "The original receipt records a build environment that is not the public replay matrix; CI therefore validates bounded semantic replay rather than bit-identical weights.", | |
| "The intrinsic nearest-neighbor checks are not a downstream benchmark.", | |
| "The kernel suite is advisory. Lambda remains Conjecture 1 (open).", | |
| "A signature or hash proves integrity and origin, not accuracy or performance." | |
| ] | |
| } | |