Upload config

config.json

@@ -1,13 +1,14 @@
 {
-  "architectures": [
-    "BertForSequenceClassification"
-  ],
+  "affine": false,
+  "alpha": 1,
   "attention_probs_dropout_prob": 0.1,
   "auto_map": {
-    "AutoConfig": "configuration_bert.BertConfig",
-    "AutoModelForSequenceClassification": "modeling_bert.BertForSequenceClassification"
+    "AutoConfig": "configuration_bert.BertConfig"
   },
+  "center": false,
   "classifier_dropout": null,
+  "do_pcr": true,
+  "do_pfsa": true,
   "hidden_act": "gelu",
   "hidden_dropout_prob": 0.1,
   "hidden_size": 768,
@@ -16,11 +17,16 @@
   "layer_norm_eps": 1e-12,
   "max_position_embeddings": 512,
   "model_type": "bert",
+  "normalize": false,
   "num_attention_heads": 12,
   "num_hidden_layers": 12,
+  "num_iters": 1,
   "pad_token_id": 0,
   "position_embedding_type": "absolute",
-  "torch_dtype": "float32",
+  "q": 5,
+  "r": 1,
+  "return_mean": true,
+  "return_std": true,
   "transformers_version": "4.33.3",
   "type_vocab_size": 2,
   "use_cache": true,

configuration_bert.py

@@ -22,7 +22,18 @@ class BertConfig(PretrainedConfig):
         pad_token_id=0,
         position_embedding_type="absolute",
         use_cache=True,
-        classifier_dropout=None,
+        classifier_dropout=None, 
+        affine=False, 
+        q=5, 
+        r=1, 
+        center=False, 
+        num_iters=1, 
+        return_mean=True, 
+        return_std=True, 
+        normalize=False, 
+        do_pcr=True, 
+        do_pfsa=True, 
+        alpha=1, 
         **kwargs,
     ):
         super().__init__(pad_token_id=pad_token_id, **kwargs)
@@ -41,4 +52,16 @@ class BertConfig(PretrainedConfig):
         self.layer_norm_eps = layer_norm_eps
         self.position_embedding_type = position_embedding_type
         self.use_cache = use_cache
-        self.classifier_dropout = classifier_dropout
+        self.classifier_dropout = classifier_dropout
+        self.affine = affine
+
+        self.q = q
+        self.r = r 
+        self.center = center
+        self.num_iters = num_iters
+        self.return_mean = return_mean
+        self.return_std = return_std
+        self.normalize = normalize
+        self.do_pcr = do_pcr
+        self.do_pfsa = do_pfsa
+        self.alpha = alpha