Text Classification
Transformers
Safetensors
modernbert
prompt-injection
jailbreak
security
multi-label
llm-guard
encoder
text-embeddings-inference
Instructions to use Accuknoxtechnologies/PromptInjection-Encoder-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use Accuknoxtechnologies/PromptInjection-Encoder-v1 with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-classification", model="Accuknoxtechnologies/PromptInjection-Encoder-v1")# Load model directly from transformers import AutoTokenizer, AutoModelForSequenceClassification tokenizer = AutoTokenizer.from_pretrained("Accuknoxtechnologies/PromptInjection-Encoder-v1") model = AutoModelForSequenceClassification.from_pretrained("Accuknoxtechnologies/PromptInjection-Encoder-v1") - Notebooks
- Google Colab
- Kaggle
add eval metrics json
Browse files- eval_metrics.json +40 -0
eval_metrics.json
ADDED
|
@@ -0,0 +1,40 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"n": 500,
|
| 3 |
+
"calibrated": true,
|
| 4 |
+
"threshold": "per-class",
|
| 5 |
+
"is_valid_threshold": 0.05,
|
| 6 |
+
"category_thresholds": {
|
| 7 |
+
"DirectInjection": 0.55,
|
| 8 |
+
"Jailbreak": 0.05,
|
| 9 |
+
"Adversarial": 0.45,
|
| 10 |
+
"Extraction": 0.55,
|
| 11 |
+
"Encoding": 0.45,
|
| 12 |
+
"Manipulation": 0.25,
|
| 13 |
+
"Smuggling": 0.65,
|
| 14 |
+
"Indirect": 0.25,
|
| 15 |
+
"MultiTurn": 0.7
|
| 16 |
+
},
|
| 17 |
+
"max_seq_length": 3072,
|
| 18 |
+
"is_valid_accuracy": 0.968,
|
| 19 |
+
"category_set_accuracy": 0.688,
|
| 20 |
+
"micro_f1": 0.7893805309734513,
|
| 21 |
+
"macro_f1": 0.7848505189708921,
|
| 22 |
+
"per_category_f1": {
|
| 23 |
+
"DirectInjection": 0.8235294117647058,
|
| 24 |
+
"Jailbreak": 0.7368421052631579,
|
| 25 |
+
"Adversarial": 0.855072463768116,
|
| 26 |
+
"Extraction": 0.7652173913043478,
|
| 27 |
+
"Encoding": 0.7516778523489933,
|
| 28 |
+
"Manipulation": 0.6785714285714286,
|
| 29 |
+
"Smuggling": 0.9256198347107438,
|
| 30 |
+
"Indirect": 0.8382352941176471,
|
| 31 |
+
"MultiTurn": 0.6888888888888889
|
| 32 |
+
},
|
| 33 |
+
"latency_ms_per_example": {
|
| 34 |
+
"mean": 1.7930222675204277,
|
| 35 |
+
"p95": 1.8397919833660126,
|
| 36 |
+
"device": "cuda:0"
|
| 37 |
+
},
|
| 38 |
+
"base_model": "jhu-clsp/mmBERT-base",
|
| 39 |
+
"epochs": 10
|
| 40 |
+
}
|