Update README.md
Browse files
README.md
CHANGED
|
@@ -11,4 +11,14 @@ tags:
|
|
| 11 |
- Guardrail
|
| 12 |
- Tool
|
| 13 |
- Classifier
|
| 14 |
-
---
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
- Guardrail
|
| 12 |
- Tool
|
| 13 |
- Classifier
|
| 14 |
+
---
|
| 15 |
+
|
| 16 |
+
|
| 17 |
+
# Tool Response Injection Detector
|
| 18 |
+
ModernBert fine-tuned over [rgeada/tool_response_injections](https://huggingface.co/datasets/rgeada/tool_response_injections). Intended for use with
|
| 19 |
+
NeMo Guardrails' [HuggingFace Classifier rails](https://github.com/NVIDIA-NeMo/Guardrails/blob/develop/docs/configure-rails/guardrail-catalog/community/hf-classifier.mdx), as a tool-response injection guardrail.
|
| 20 |
+
|
| 21 |
+
|
| 22 |
+
## Scores:
|
| 23 |
+
- Test F1-Score: 0.8464
|
| 24 |
+
- Test Accuracy: 0.8466
|