|
|
--- |
|
|
base_model: |
|
|
- protectai/deberta-v3-base-prompt-injection |
|
|
pipeline_tag: text-classification |
|
|
language: |
|
|
- en |
|
|
tags: |
|
|
- prompt-injection |
|
|
- injection |
|
|
- security |
|
|
- generated_from_trainer |
|
|
datasets: |
|
|
- xTRam1/safe-guard-prompt-injection |
|
|
metrics: |
|
|
- accuracy |
|
|
- recall |
|
|
- precision |
|
|
- f1 |
|
|
--- |
|
|
|
|
|
# Model Card for bigberta-v1-pompt-injection |
|
|
|
|
|
This model is a fine-tuned version of [protectai/deberta-v3-base-prompt-injection](https://huggingface.co/protectai/deberta-v3-base-prompt-injection) on multiple datasets of prompt injections. |
|
|
|
|
|
It aims to identify prompt injections, classifying inputs into two categories: `0` for no injection and `1` for injection detected. |
|
|
|
|
|
It achieves the following results on the evaluation set: |
|
|
|
|
|
Test Samples: 2060 |
|
|
|
|
|
- Loss: 0.0361 |
|
|
- Accuracy: 0.9908 |
|
|
- Precision: 0.9861 |
|
|
- Recall: 0.9846 |
|
|
- F1 Score: 0.9854 |
