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
Release CLEF cybersecurity detector with Jev and Laya R2a comparison
Browse files- README.md +3 -1
- assets/benchmark-auroc.svg +44 -44
- assets/benchmark-pdfs.svg +41 -41
- release_manifest.json +4 -4
README.md
CHANGED
|
@@ -84,6 +84,8 @@ These are **previously inspected internal regression sets**, not a public leader
|
|
| 84 |
|
| 85 |
Use a CUDA GPU with enough memory for the full base. The measured batch-one runtime allocated about **19.8 GiB**; allow additional VRAM headroom. Longer documents and larger batches need more memory.
|
| 86 |
|
|
|
|
|
|
|
| 87 |
```bash
|
| 88 |
pip install -r requirements.txt
|
| 89 |
```
|
|
@@ -131,4 +133,4 @@ These timings are not a matched latency comparison against the hosted Jev API or
|
|
| 131 |
|
| 132 |
The inference definitions are exported from the tested training/evaluation implementation. The loader pins the public base revision, verifies its native Python-source hash before import, and retains trained FP32 parameter precision. [release_manifest.json](release_manifest.json) records file hashes, model size, and source snapshot digests.
|
| 133 |
|
| 134 |
-
Released under **Apache-2.0**; see [LICENSE](LICENSE) and [NOTICE](NOTICE). Credit: [Cloudflare CLEF](https://huggingface.co/Cloudflare/clef-flash), [Qwen3.5-9B](https://huggingface.co/Qwen/Qwen3.5-9B), and TextCortex's domain fine-tuning and evaluation work. This repository is not an official Cloudflare release or endorsement.
|
|
|
|
| 84 |
|
| 85 |
Use a CUDA GPU with enough memory for the full base. The measured batch-one runtime allocated about **19.8 GiB**; allow additional VRAM headroom. Longer documents and larger batches need more memory.
|
| 86 |
|
| 87 |
+
Download this repository, then install the pinned runtime requirements from its directory:
|
| 88 |
+
|
| 89 |
```bash
|
| 90 |
pip install -r requirements.txt
|
| 91 |
```
|
|
|
|
| 133 |
|
| 134 |
The inference definitions are exported from the tested training/evaluation implementation. The loader pins the public base revision, verifies its native Python-source hash before import, and retains trained FP32 parameter precision. [release_manifest.json](release_manifest.json) records file hashes, model size, and source snapshot digests.
|
| 135 |
|
| 136 |
+
Released under **Apache-2.0**; see [LICENSE](https://huggingface.co/TextCortex/clef-cybersecurity/blob/main/LICENSE) and [NOTICE](https://huggingface.co/TextCortex/clef-cybersecurity/blob/main/NOTICE). Credit: [Cloudflare CLEF](https://huggingface.co/Cloudflare/clef-flash), [Qwen3.5-9B](https://huggingface.co/Qwen/Qwen3.5-9B), and TextCortex's domain fine-tuning and evaluation work. This repository is not an official Cloudflare release or endorsement.
|
assets/benchmark-auroc.svg
CHANGED
|
|
|
|
assets/benchmark-pdfs.svg
CHANGED
|
|
|
|
release_manifest.json
CHANGED
|
@@ -33,8 +33,8 @@
|
|
| 33 |
"sha256": "e9954b5e3ea46a528aeac8643a36aa209d84e9790019307303fc7ed01a794e98"
|
| 34 |
},
|
| 35 |
"README.md": {
|
| 36 |
-
"bytes":
|
| 37 |
-
"sha256": "
|
| 38 |
},
|
| 39 |
"clef_detector.py": {
|
| 40 |
"bytes": 16866,
|
|
@@ -46,11 +46,11 @@
|
|
| 46 |
},
|
| 47 |
"assets/benchmark-pdfs.svg": {
|
| 48 |
"bytes": 64179,
|
| 49 |
-
"sha256": "
|
| 50 |
},
|
| 51 |
"assets/benchmark-auroc.svg": {
|
| 52 |
"bytes": 75256,
|
| 53 |
-
"sha256": "
|
| 54 |
},
|
| 55 |
"assets/benchmark-auroc.png": {
|
| 56 |
"bytes": 131047,
|
|
|
|
| 33 |
"sha256": "e9954b5e3ea46a528aeac8643a36aa209d84e9790019307303fc7ed01a794e98"
|
| 34 |
},
|
| 35 |
"README.md": {
|
| 36 |
+
"bytes": 9726,
|
| 37 |
+
"sha256": "6ae456f24bd0674761ff7209412736394885f411538686e556c18262b90b44ba"
|
| 38 |
},
|
| 39 |
"clef_detector.py": {
|
| 40 |
"bytes": 16866,
|
|
|
|
| 46 |
},
|
| 47 |
"assets/benchmark-pdfs.svg": {
|
| 48 |
"bytes": 64179,
|
| 49 |
+
"sha256": "b4433e9f664d377dbfabbe056adad3d5b7d3422807299a3893aafcae658b7a5c"
|
| 50 |
},
|
| 51 |
"assets/benchmark-auroc.svg": {
|
| 52 |
"bytes": 75256,
|
| 53 |
+
"sha256": "d568614a7245d65746f6b31040fb5259a15b0b16f10d456f0312b61a7392fe79"
|
| 54 |
},
|
| 55 |
"assets/benchmark-auroc.png": {
|
| 56 |
"bytes": 131047,
|