cgoosen/Prompt-Guard-finetuned-ctf-86M

README.md

@@ -1,6 +1,6 @@
 ---
 license: mit
-base_model: microsoft/mdeberta-v3-base
+base_model: microsoft/deberta-v3-base
 tags:
 - generated_from_trainer
 metrics:
@@ -15,10 +15,10 @@ should probably proofread and complete it, then remove this comment. -->
 
 # Prompt-Guard-finetuned-ctf-86M
 
-This model is a fine-tuned version of [microsoft/mdeberta-v3-base](https://huggingface.co/microsoft/mdeberta-v3-base) on the None dataset.
+This model is a fine-tuned version of [microsoft/deberta-v3-base](https://huggingface.co/microsoft/deberta-v3-base) on the None dataset.
 It achieves the following results on the evaluation set:
-- Loss: 0.1377
-- Accuracy: 0.9376
+- Loss: 0.0226
+- Accuracy: 0.9964
 
 ## Model description
 
@@ -38,27 +38,23 @@ More information needed
 
 The following hyperparameters were used during training:
 - learning_rate: 2e-05
-- train_batch_size: 16
-- eval_batch_size: 16
+- train_batch_size: 8
+- eval_batch_size: 8
 - seed: 42
 - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
 - lr_scheduler_type: linear
-- num_epochs: 10
+- num_epochs: 6
 
 ### Training results
 
-| Training Loss | Epoch | Step | Validation Loss | Accuracy |
-|:-------------:|:-----:|:----:|:---------------:|:--------:|
-| 0.2424        | 1.0   | 293  | 0.1377          | 0.9376   |
-| 0.1436        | 2.0   | 586  | 0.1565          | 0.9487   |
-| 0.108         | 3.0   | 879  | 0.2244          | 0.9462   |
-| 0.0806        | 4.0   | 1172 | 0.2356          | 0.9462   |
-| 0.0548        | 5.0   | 1465 | 0.2658          | 0.9470   |
-| 0.0473        | 6.0   | 1758 | 0.2437          | 0.9521   |
-| 0.0339        | 7.0   | 2051 | 0.2838          | 0.9470   |
-| 0.0199        | 8.0   | 2344 | 0.2974          | 0.9530   |
-| 0.0193        | 9.0   | 2637 | 0.3392          | 0.9487   |
-| 0.0176        | 10.0  | 2930 | 0.3296          | 0.9513   |
+| Training Loss | Epoch | Step  | Validation Loss | Accuracy |
+|:-------------:|:-----:|:-----:|:---------------:|:--------:|
+| 0.0291        | 1.0   | 2346  | 0.0315          | 0.9953   |
+| 0.0222        | 2.0   | 4692  | 0.0318          | 0.9953   |
+| 0.0081        | 3.0   | 7038  | 0.0226          | 0.9964   |
+| 0.0032        | 4.0   | 9384  | 0.0329          | 0.9966   |
+| 0.0031        | 5.0   | 11730 | 0.0352          | 0.9957   |
+| 0.0039        | 6.0   | 14076 | 0.0242          | 0.9974   |
 
 
 ### Framework versions

added_tokens.json

@@ -1,3 +1,3 @@
 {
-  "[MASK]": 250101
+  "[MASK]": 128000
 }

config.json

@@ -1,5 +1,5 @@
 {
-  "_name_or_path": "microsoft/mdeberta-v3-base",
+  "_name_or_path": "microsoft/deberta-v3-base",
   "architectures": [
     "DebertaV2ForSequenceClassification"
   ],
@@ -8,14 +8,16 @@
   "hidden_dropout_prob": 0.1,
   "hidden_size": 768,
   "id2label": {
-    "0": "POSITIVE",
-    "1": "NEGATIVE"
+    "0": "benign",
+    "1": "injection",
+    "2": "jailbreak"
   },
   "initializer_range": 0.02,
   "intermediate_size": 3072,
   "label2id": {
-    "NEGATIVE": 1,
-    "POSITIVE": 0
+    "benign": 0,
+    "injection": 1,
+    "jailbreak": 2
   },
   "layer_norm_eps": 1e-07,
   "max_position_embeddings": 512,
@@ -39,5 +41,5 @@
   "torch_dtype": "float32",
   "transformers_version": "4.40.2",
   "type_vocab_size": 0,
-  "vocab_size": 251000
+  "vocab_size": 128100
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:08fee00422e08c2fbe00beb31889ad4be003b073ebeef090112432c155502cf0
-size 1115268200
+oid sha256:078563f4d4da764b13e136573021c0dfe14dbb3c168665d23a895b08e442c6cd
+size 737722356

spm.model

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:13c8d666d62a7bc4ac8f040aab68e942c861f93303156cc28f5c7e885d86d6e3
-size 4305025
+oid sha256:c679fbf93643d19aab7ee10c0b99e460bdbc02fedf34b92b05af343b4af586fd
+size 2464616

tokenizer.json

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5e26ec8cd2234df8edcc6e8c207b66b3162360f9fdc46ee4526a440d0d9d0bb9
-size 16331374
+oid sha256:5124ef2ead1a10a717703bc436de7f353da76d6340e4587719b42b1693707964
+size 8656624

tokenizer_config.json

@@ -32,7 +32,7 @@
       "single_word": false,
       "special": true
     },
-    "250101": {
+    "128000": {
       "content": "[MASK]",
       "lstrip": false,
       "normalized": false,

training_args.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:adea9aef635495a08f557a4ac41e5232bf934163c31053be61c99a68480b4fc7
+oid sha256:38f9ca1e0d7904a29e057a2ece9bba73972fd0b1667a3bb847c67cde6d2061d2
 size 5048