test

README.md

@@ -40,7 +40,12 @@ Please read our technical paper to get the detailed performance comparison here:
 
 ## Usage Example
 
-You can use these models directly with the Hugging Face Transformers library for classification tasks. Below is an example to classify a prompt as malicious or benign:
+You can use these models directly with the Hugging Face Transformers library for classification tasks. Below is an example to classify a prompt as malicious or benign.
+
+First install dependencies:
+```
+pip install transformers torch huggingface-hub tokenizers sentencepiece safetensors protobuf
+```
 
 ```python
 from transformers import AutoTokenizer, AutoModelForSequenceClassification, pipeline
@@ -78,4 +83,4 @@ The models have been evaluated across multiple datasets:
 - [Microsoft-BIPIA](https://github.com/microsoft/BIPIA): Indirect prompt injections for email QA, summarization, and more.
 - [JailbreakBench](https://jailbreakbench.github.io/): JBB-Behaviors artifacts composed of 100 distinct misuse behaviors.
 - [Garak Vulnerability Scanner](https://github.com/NVIDIA/garak): Red-teaming assessments with diverse attack types.
-- Real-World Attacks: Benchmarked against real-world malicious prompts.
+- Real-World Attacks: Benchmarked against real-world malicious prompts.

config.json

@@ -1,22 +1,21 @@
 {
-  "_name_or_path": "microsoft/deberta-v3-small",
   "architectures": [
     "DebertaV2ForSequenceClassification"
   ],
-    "label2id": {
-    "INJECTION": 1,
-    "SAFE": 0
-  },
-  "id2label": {
-    "0": "SAFE",
-    "1": "INJECTION"
-  },
   "attention_probs_dropout_prob": 0.1,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
   "hidden_size": 768,
+  "id2label": {
+    "0": "SAFE",
+    "1": "INJECTION"
+  },
   "initializer_range": 0.02,
   "intermediate_size": 3072,
+  "label2id": {
+    "INJECTION": 1,
+    "SAFE": 0
+  },
   "layer_norm_eps": 1e-07,
   "legacy": true,
   "max_position_embeddings": 512,
@@ -38,7 +37,7 @@
   "relative_attention": true,
   "share_att_key": true,
   "torch_dtype": "float32",
-  "transformers_version": "4.48.3",
+  "transformers_version": "4.55.4",
   "type_vocab_size": 0,
   "vocab_size": 128100
 }

model.onnx → onnx/model.onnx

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:a55639b3619c2a0d2e0b4405dc35a341e4707f0ed774c4893f146c84412b79ff
-size 568071686
+oid sha256:a70600837172031b9b5ec35c7b2e6f15ed4e91b9a5c6137d43df679fa09a4be5
+size 568046172

special_tokens_map.json

@@ -1,10 +1,46 @@
 {
-  "bos_token": "[CLS]",
-  "cls_token": "[CLS]",
-  "eos_token": "[SEP]",
-  "mask_token": "[MASK]",
-  "pad_token": "[PAD]",
-  "sep_token": "[SEP]",
+  "bos_token": {
+    "content": "[CLS]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "cls_token": {
+    "content": "[CLS]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "eos_token": {
+    "content": "[SEP]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "mask_token": {
+    "content": "[MASK]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "pad_token": {
+    "content": "[PAD]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
+  "sep_token": {
+    "content": "[SEP]",
+    "lstrip": false,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  },
   "unk_token": {
     "content": "[UNK]",
     "lstrip": false,