Upload folder using huggingface_hub

README.md

@@ -16,22 +16,22 @@ tags:
 - coreference-resolution
 - distilbert
 - multi-task
-base_model: distilbert-base-cased
+base_model: microsoft/deberta-v3-small
 ---
 
 # Kiji PII Detection Model
 
-Multi-task DistilBERT model for detecting Personally Identifiable Information (PII) in text with coreference resolution. Fine-tuned from [`distilbert-base-cased`](https://huggingface.co/distilbert-base-cased).
+Multi-task DistilBERT model for detecting Personally Identifiable Information (PII) in text with coreference resolution. Fine-tuned from [`microsoft/deberta-v3-small`](https://huggingface.co/microsoft/deberta-v3-small).
 
 ## Model Summary
 
 | | |
 |---|---|
-| **Base model** | [distilbert-base-cased](https://huggingface.co/distilbert-base-cased) |
+| **Base model** | [microsoft/deberta-v3-small](https://huggingface.co/microsoft/deberta-v3-small) |
 | **Architecture** | Shared DistilBERT encoder + two linear classification heads |
 | **Parameters** | ~66M |
-| **Model size** | 249 MB (SafeTensors) |
-| **Tasks** | PII token classification (53 labels) + coreference detection (7 labels) |
+| **Model size** | 703 MB (SafeTensors) |
+| **Tasks** | PII token classification (53 labels) + coreference detection (0 labels) |
 | **PII entity types** | 26 |
 | **Max sequence length** | 512 tokens |
 
@@ -45,7 +45,7 @@ Input (input_ids, attention_mask)
    +----+----+
    |         |
 PII Head  Coref Head
-(768->53)  (768->7)
+(768->53)  (768->0)
 ```
 
 The model uses multi-task learning: a shared DistilBERT encoder feeds into two independent linear classification heads. Both tasks are trained simultaneously with equal loss weighting, which acts as regularization and improves PII detection generalization.
@@ -132,12 +132,6 @@ Each entity type has `B-` (beginning) and `I-` (inside) variants, plus `O` for n
 
 Trained on the [DataikuNLP/kiji-pii-training-data](https://huggingface.co/datasets/DataikuNLP/kiji-pii-training-data) dataset — a synthetic multilingual PII dataset with entity annotations and coreference resolution.
 
-## Derived Models
-
-| Variant | Format | Repository |
-|---------|--------|------------|
-| Quantized (INT8) | ONNX | [DataikuNLP/kiji-pii-model-onnx](https://huggingface.co/DataikuNLP/kiji-pii-model-onnx) |
-
 ## Limitations
 
 - Trained on **synthetically generated** data — may not generalize to all real-world text

added_tokens.json

@@ -0,0 +1,3 @@
+{
+  "[MASK]": 128000
+}

config.json

@@ -0,0 +1,33 @@
+{
+  "_attn_implementation_autoset": true,
+  "attention_probs_dropout_prob": 0.1,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-07,
+  "legacy": true,
+  "max_position_embeddings": 512,
+  "max_relative_positions": -1,
+  "model_type": "deberta-v2",
+  "norm_rel_ebd": "layer_norm",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 0,
+  "pooler_dropout": 0,
+  "pooler_hidden_act": "gelu",
+  "pooler_hidden_size": 768,
+  "pos_att_type": [
+    "p2c",
+    "c2p"
+  ],
+  "position_biased_input": false,
+  "position_buckets": 256,
+  "relative_attention": true,
+  "share_att_key": true,
+  "torch_dtype": "float32",
+  "transformers_version": "4.51.3",
+  "type_vocab_size": 0,
+  "vocab_size": 128100
+}

label_mappings.json

@@ -111,16 +111,5 @@
       "52": "I-USERNAME",
       "-100": "IGNORE"
     }
-  },
-  "coref": {
-    "id2label": {
-      "0": "NO_COREF",
-      "1": "CLUSTER_0",
-      "2": "CLUSTER_1",
-      "3": "CLUSTER_2",
-      "4": "CLUSTER_3",
-      "5": "CLUSTER_4",
-      "6": "CLUSTER_5"
-    }
   }
 }

model.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:246b7e3f3f1e0369155ad84c55efa9769ddd149861f9ce7f93a8f293ab58ee7e
-size 260960440
+oid sha256:3d41bd69ac00fbf7bb546870200feb2a6d5df4c2b04150acd35ce4b0fe8ad6e1
+size 736625504

special_tokens_map.json

@@ -1,7 +1,15 @@
 {
+  "bos_token": "[CLS]",
   "cls_token": "[CLS]",
+  "eos_token": "[SEP]",
   "mask_token": "[MASK]",
   "pad_token": "[PAD]",
   "sep_token": "[SEP]",
-  "unk_token": "[UNK]"
+  "unk_token": {
+    "content": "[UNK]",
+    "lstrip": false,
+    "normalized": true,
+    "rstrip": false,
+    "single_word": false
+  }
 }

spm.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:c679fbf93643d19aab7ee10c0b99e460bdbc02fedf34b92b05af343b4af586fd
+size 2464616

tokenizer_config.json

@@ -8,31 +8,31 @@
       "single_word": false,
       "special": true
     },
-    "100": {
-      "content": "[UNK]",
+    "1": {
+      "content": "[CLS]",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "101": {
-      "content": "[CLS]",
+    "2": {
+      "content": "[SEP]",
       "lstrip": false,
       "normalized": false,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "102": {
-      "content": "[SEP]",
+    "3": {
+      "content": "[UNK]",
       "lstrip": false,
-      "normalized": false,
+      "normalized": true,
       "rstrip": false,
       "single_word": false,
       "special": true
     },
-    "103": {
+    "128000": {
       "content": "[MASK]",
       "lstrip": false,
       "normalized": false,
@@ -41,16 +41,19 @@
       "special": true
     }
   },
+  "bos_token": "[CLS]",
   "clean_up_tokenization_spaces": false,
   "cls_token": "[CLS]",
   "do_lower_case": false,
+  "eos_token": "[SEP]",
   "extra_special_tokens": {},
   "mask_token": "[MASK]",
-  "model_max_length": 512,
+  "model_max_length": 1000000000000000019884624838656,
   "pad_token": "[PAD]",
   "sep_token": "[SEP]",
-  "strip_accents": null,
-  "tokenize_chinese_chars": true,
-  "tokenizer_class": "DistilBertTokenizer",
-  "unk_token": "[UNK]"
+  "sp_model_kwargs": {},
+  "split_by_punct": false,
+  "tokenizer_class": "DebertaV2Tokenizer",
+  "unk_token": "[UNK]",
+  "vocab_type": "spm"
 }