Text Classification
Transformers
Safetensors
English
German
cybersecurity
prompt-injection
data-exfiltration
clef
custom-code
Eval Results (legacy)
Instructions to use TextCortex/clef-cybersecurity with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use TextCortex/clef-cybersecurity with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="TextCortex/clef-cybersecurity")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("TextCortex/clef-cybersecurity", device_map="auto") - Notebooks
- Google Colab
- Kaggle
Download benchmarks.json from TextCortex/clef-cybersecurity: direct link, hf CLI and curl.
- Browser
- Download file 3.56 kB
-
https://huggingface.co/TextCortex/clef-cybersecurity/resolve/main/benchmarks.json
- Command line
-
hf download hf://TextCortex/clef-cybersecurity/benchmarks.json
-
curl -L -o benchmarks.json https://huggingface.co/TextCortex/clef-cybersecurity/resolve/main/benchmarks.json
3.56 kB
| { | |
| "type": "aggregate_prompt_injection_benchmarks", | |
| "models": [ | |
| { | |
| "model": "TextCortex/clef-cybersecurity", | |
| "name": "clef-cybersecurity", | |
| "full_en": { | |
| "n": 510, | |
| "attacks": 256, | |
| "clean": 254, | |
| "auroc": 0.9925335260826772, | |
| "caught": 232, | |
| "false_alarms": 0 | |
| }, | |
| "full_de": { | |
| "n": 510, | |
| "attacks": 256, | |
| "clean": 254, | |
| "auroc": 0.9743940698818898, | |
| "caught": 241, | |
| "false_alarms": 17 | |
| }, | |
| "skills_en": { | |
| "n": 48, | |
| "attacks": 27, | |
| "clean": 21, | |
| "auroc": 1.0, | |
| "caught": 23, | |
| "false_alarms": 0 | |
| }, | |
| "skills_de": { | |
| "n": 48, | |
| "attacks": 27, | |
| "clean": 21, | |
| "auroc": 0.9171075837742504, | |
| "caught": 26, | |
| "false_alarms": 3 | |
| }, | |
| "pdf": { | |
| "n": 730, | |
| "attacks": 107, | |
| "clean": 623, | |
| "auroc": 0.9856212778086137, | |
| "caught": 84, | |
| "false_alarms": 3 | |
| }, | |
| "decision_rule": "score > 0.5" | |
| }, | |
| { | |
| "model": "jev", | |
| "name": "Jev", | |
| "full_en": { | |
| "auroc": 0.9799920029527559, | |
| "n": 510 | |
| }, | |
| "full_de": { | |
| "auroc": 0.9564391609251969, | |
| "n": 510 | |
| }, | |
| "skills_en": { | |
| "n": 48, | |
| "attacks": 27, | |
| "clean": 21, | |
| "auroc": 0.9841269841269841, | |
| "caught": 25, | |
| "false_alarms": 1 | |
| }, | |
| "skills_de": { | |
| "n": 48, | |
| "attacks": 27, | |
| "clean": 21, | |
| "auroc": 0.9603174603174603, | |
| "caught": 25, | |
| "false_alarms": 2 | |
| }, | |
| "pdf": { | |
| "n": 730, | |
| "attacks": 107, | |
| "clean": 623, | |
| "auroc": 0.978525674682348, | |
| "caught": 73, | |
| "false_alarms": 2 | |
| }, | |
| "decision_rule": "score > 0.5" | |
| }, | |
| { | |
| "model": "laya-cybersec-r2a", | |
| "name": "Laya R2a", | |
| "full_en": { | |
| "n": 510, | |
| "attacks": 256, | |
| "clean": 254, | |
| "auroc": 0.9155004306102362, | |
| "caught": 140, | |
| "false_alarms": 2, | |
| "recall": 0.546875, | |
| "fpr": 0.007874015748031496 | |
| }, | |
| "full_de": { | |
| "n": 510, | |
| "attacks": 256, | |
| "clean": 254, | |
| "auroc": 0.8780065821850394, | |
| "caught": 151, | |
| "false_alarms": 9, | |
| "recall": 0.58984375, | |
| "fpr": 0.03543307086614173 | |
| }, | |
| "skills_en": { | |
| "n": 48, | |
| "attacks": 27, | |
| "clean": 21, | |
| "auroc": 0.927689594356261, | |
| "caught": 18, | |
| "false_alarms": 0 | |
| }, | |
| "skills_de": { | |
| "n": 48, | |
| "attacks": 27, | |
| "clean": 21, | |
| "auroc": 0.9329805996472663, | |
| "caught": 18, | |
| "false_alarms": 0 | |
| }, | |
| "pdf": { | |
| "n": 730, | |
| "attacks": 107, | |
| "clean": 623, | |
| "auroc": 0.8855927753859077, | |
| "caught": 81, | |
| "false_alarms": 0 | |
| }, | |
| "decision_rule": "score > 0.95" | |
| } | |
| ], | |
| "cohort": { | |
| "full_per_language": 510, | |
| "skills_per_language": 48, | |
| "pdf_attacks": 107, | |
| "clean_pdfs": 623 | |
| }, | |
| "scope": "Previously inspected internal regression sets; not a public leaderboard or fresh blind test.", | |
| "threshold_note": "Strict >0.5 for CLEF and Jev; strict >0.95 for Laya R2a. Recall and false alarms are not equal-threshold comparisons.", | |
| "selected_epoch": 3, | |
| "epochs_trained": 4, | |
| "clef_secondary_pdf_at_095": { | |
| "n": 730, | |
| "attacks": 107, | |
| "clean": 623, | |
| "auroc": 0.9856212778086137, | |
| "caught": 69, | |
| "false_alarms": 2 | |
| } | |
| } | |