Sentence Similarity
sentence-transformers
Joblib
Safetensors
modernbert
security
intrusion-detection
behavior-analytics
intent-recognition
linux
kubernetes
audit-log
text-embeddings-inference
Instructions to use willchen0011/SecEBL with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- sentence-transformers
How to use willchen0011/SecEBL with sentence-transformers:
from sentence_transformers import SentenceTransformer model = SentenceTransformer("willchen0011/SecEBL") sentences = [ "That is a happy person", "That is a happy dog", "That is a very happy person", "Today is a sunny day" ] embeddings = model.encode(sentences) similarities = model.similarity(embeddings, embeddings) print(similarities.shape) # [4, 4] - Notebooks
- Google Colab
- Kaggle
| { | |
| "schema": "secebl_rev20_l2_public_summary_v1", | |
| "l1_release": "SecEBL-Rev20 public L1 release", | |
| "l2_release": "SecEBL-Rev20 public L2 artifact", | |
| "model_type": "logistic_regression_session_scorer", | |
| "scoring_scope": "experimental fitted L2 session scorer over cached L1 semantic features", | |
| "runtime_feature_exclusions": [ | |
| "raw_command_text", | |
| "user_name", | |
| "host_name", | |
| "session_id" | |
| ], | |
| "training_mixture": { | |
| "sessions": 5747, | |
| "positive_sessions": 426, | |
| "negative_sessions": 5321, | |
| "synthetic_pressure_positive_sessions": 60, | |
| "reviewed_real_pressure_positive_sessions": 1, | |
| "random_real_pressure_background_negative_sessions": 5000, | |
| "reviewed_hard_negative_pressure_sessions": 23 | |
| }, | |
| "validation_oof": { | |
| "folds": 5, | |
| "accuracy": 0.9939098660170523, | |
| "attack_precision": 0.9643705463182898, | |
| "attack_recall": 0.9530516431924883, | |
| "normal_recall": 0.9971809810186055, | |
| "tp": 406, | |
| "fn": 20, | |
| "fp": 15, | |
| "tn": 5306 | |
| }, | |
| "withheld_session_benchmark_fit_check": { | |
| "sessions": 663, | |
| "rows_seen": 12594, | |
| "attack_sessions": 365, | |
| "normal_sessions": 298, | |
| "accuracy": 1.0, | |
| "attack_precision": 1.0, | |
| "attack_recall": 1.0, | |
| "normal_recall": 1.0, | |
| "tp": 365, | |
| "fn": 0, | |
| "fp": 0, | |
| "tn": 298 | |
| }, | |
| "pressure_stream_fit_check": { | |
| "rows_seen": 6286568, | |
| "sessions": 102117, | |
| "alert_sessions": 61, | |
| "reviewed_real_alert_sessions": 1, | |
| "synthetic_alert_sessions": 60, | |
| "raw_rows_redistributed": false, | |
| "real_session_identifiers_redistributed": false | |
| }, | |
| "thresholds": { | |
| "score_threshold": 0.5, | |
| "model_probability_threshold": 0.9, | |
| "score_transform": "threshold_margin", | |
| "score_transform_scale": 2.0 | |
| }, | |
| "caveat": "This is an experimental fitted L2 artifact for reproducible SecEBL-Rev20 session experiments, not an independent claim of general production IDS accuracy." | |
| } | |