Upload README.md with huggingface_hub

README.md

@@ -0,0 +1,59 @@
+---
+license: mit
+language:
+- en
+datasets:
+- clawdia-chan/prompt-guard-v1
+---
+
+# Balanced DeBERTa Prompt Guard
+
+Fine-tuned DeBERTa-v3-base for prompt injection and jailbreak detection.
+
+## Model Details
+
+- **Base Model:** microsoft/deberta-v3-base
+- **Training Data:** 500+ balanced samples (prompt-guard-v1 dataset)
+- **Task:** Binary classification (malicious vs benign prompts)
+
+## Supported Threats
+
+| Threat Type | Status |
+|-------------|--------|
+| Direct Prompt Injection | ✅ |
+| Indirect Injection | ✅ |
+| FlipAttack | ✅ |
+| Jailbreak (DAN) | ✅ |
+| Encoding Attacks | ✅ |
+| Role Override | ✅ |
+
+## Performance
+
+- Training accuracy: ~100% (small test set)
+- Validation: Manual testing on real attack patterns
+
+## Usage
+
+```python
+from transformers import AutoModelForSequenceClassification, AutoTokenizer
+
+model_name = "clawdia-chan/balanced-deberta"
+tokenizer = AutoTokenizer.from_pretrained(model_name)
+model = AutoModelForSequenceClassification.from_pretrained(model_name)
+```
+
+## Limitations
+
+- Small training dataset (500+ samples)
+- No formal evaluation metrics published
+- Requires more diverse test data
+
+## Citation
+
+```
+@model{phantomshield,
+  title={Balanced DeBERTa Prompt Guard},
+  author={Clawdia},
+  year={2026}
+}
+```