Text Classification
Scikit-learn
Joblib
Safetensors
promptguard
privacy
dlp
korean
roberta
logistic-regression
Instructions to use OASecure/promptguard-context-classifier with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Scikit-learn
How to use OASecure/promptguard-context-classifier with Scikit-learn:
from huggingface_hub import hf_hub_download import joblib model = joblib.load( hf_hub_download("OASecure/promptguard-context-classifier", "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
Fix model card base model metadata
Browse files
README.md
CHANGED
|
@@ -1,6 +1,5 @@
|
|
| 1 |
---
|
| 2 |
license: other
|
| 3 |
-
library_name: transformers
|
| 4 |
pipeline_tag: text-classification
|
| 5 |
tags:
|
| 6 |
- promptguard
|
|
@@ -9,8 +8,10 @@ tags:
|
|
| 9 |
- dlp
|
| 10 |
- korean
|
| 11 |
- roberta
|
|
|
|
|
|
|
|
|
|
| 12 |
base_model:
|
| 13 |
-
- Qwen/Qwen3-Embedding-0.6B
|
| 14 |
- klue/roberta-base
|
| 15 |
---
|
| 16 |
|
|
@@ -20,12 +21,14 @@ This repository hosts the runtime model artifacts for PromptGuard's context-risk
|
|
| 20 |
|
| 21 |
## Runtime Architecture
|
| 22 |
|
| 23 |
-
PromptGuard uses a two-stage context classifier
|
| 24 |
|
| 25 |
-
1. Qwen embedding + One-vs-Rest Logistic Regression candidate generator
|
| 26 |
2. Fine-tuned KLUE RoBERTa label-aware verifier
|
| 27 |
|
| 28 |
-
The Logistic Regression stage is a high-recall candidate generator
|
|
|
|
|
|
|
| 29 |
|
| 30 |
## Files
|
| 31 |
|
|
@@ -80,13 +83,15 @@ PROMPTGUARD_VERIFIER_RUNTIME_ENABLED=true
|
|
| 80 |
PROMPTGUARD_VERIFIER_MANIFEST_PATH=/opt/promptguard/models/context_lr_roberta_active_best_f1_manifest.json
|
| 81 |
```
|
| 82 |
|
|
|
|
|
|
|
| 83 |
## Data and Privacy
|
| 84 |
|
| 85 |
-
This repository contains runtime artifacts only. It does not include raw prompts, source documents, extracted text, filenames,
|
| 86 |
|
| 87 |
## License Notes
|
| 88 |
|
| 89 |
-
The PromptGuard application code is distributed separately on GitHub. These model artifacts are published for PromptGuard runtime use. `Qwen/Qwen3-Embedding-0.6B`
|
| 90 |
|
| 91 |
## Limitations
|
| 92 |
|
|
|
|
| 1 |
---
|
| 2 |
license: other
|
|
|
|
| 3 |
pipeline_tag: text-classification
|
| 4 |
tags:
|
| 5 |
- promptguard
|
|
|
|
| 8 |
- dlp
|
| 9 |
- korean
|
| 10 |
- roberta
|
| 11 |
+
- sklearn
|
| 12 |
+
- logistic-regression
|
| 13 |
+
- joblib
|
| 14 |
base_model:
|
|
|
|
| 15 |
- klue/roberta-base
|
| 16 |
---
|
| 17 |
|
|
|
|
| 21 |
|
| 22 |
## Runtime Architecture
|
| 23 |
|
| 24 |
+
PromptGuard uses a two-stage context classifier. This repository is a runtime artifact package, not a single Transformer checkpoint loaded from the repository root.
|
| 25 |
|
| 26 |
+
1. Frozen Qwen embedding dependency + One-vs-Rest Logistic Regression candidate generator
|
| 27 |
2. Fine-tuned KLUE RoBERTa label-aware verifier
|
| 28 |
|
| 29 |
+
The Logistic Regression stage is a high-recall linear candidate generator stored as a `joblib` artifact. It is not a fine-tuned Qwen model and does not have its own Hugging Face base model. It expects normalized embedding vectors produced by `Qwen/Qwen3-Embedding-0.6B`.
|
| 30 |
+
|
| 31 |
+
The RoBERTa verifier is the fine-tuned Transformer stage. Its base model is `klue/roberta-base`, and the fine-tuned verifier checkpoint is stored under `models/context_verifier_klue_roberta_base_lrmined_v287_global002_compactv2_lpft_focal_1p2ep/`.
|
| 32 |
|
| 33 |
## Files
|
| 34 |
|
|
|
|
| 83 |
PROMPTGUARD_VERIFIER_MANIFEST_PATH=/opt/promptguard/models/context_lr_roberta_active_best_f1_manifest.json
|
| 84 |
```
|
| 85 |
|
| 86 |
+
Do not load this repository root directly with `AutoModelForSequenceClassification.from_pretrained("OASecure/promptguard-context-classifier")`. PromptGuard reads `models/context_lr_roberta_active_best_f1_manifest.json`, loads the LR `joblib` file for candidate generation, and loads the verifier directory named in that manifest for RoBERTa verification.
|
| 87 |
+
|
| 88 |
## Data and Privacy
|
| 89 |
|
| 90 |
+
This repository contains runtime artifacts only. It does not include raw prompts, source documents, extracted text, filenames, training/evaluation detail rows, embedding vectors, logits, exact classifier scores, or full masked prompts.
|
| 91 |
|
| 92 |
## License Notes
|
| 93 |
|
| 94 |
+
The PromptGuard application code is distributed separately on GitHub. These model artifacts are published for PromptGuard runtime use. `Qwen/Qwen3-Embedding-0.6B` is used as a frozen embedding dependency for the LR stage, not as the fine-tuned base model of this repository. `klue/roberta-base` is the verifier base model. Downstream users should review base-model and training-data obligations before redistribution or commercial use.
|
| 95 |
|
| 96 |
## Limitations
|
| 97 |
|