model improved

config.json

@@ -40,7 +40,7 @@
     "28": "B-PART",
     "29": "B-PRON",
     "30": "B-PROPN",
-    "31": "B-PUNCT",
+    "31": "B-PUNCT+NOUN",
     "32": "B-SCONJ",
     "33": "B-SYM",
     "34": "B-VERB",
@@ -77,7 +77,7 @@
     "65": "I-PART",
     "66": "I-PRON",
     "67": "I-PROPN",
-    "68": "I-PUNCT",
+    "68": "I-PUNCT+NOUN",
     "69": "I-SCONJ",
     "70": "I-SYM",
     "71": "I-VERB",
@@ -96,12 +96,14 @@
     "84": "PRON+ADP",
     "85": "PROPN",
     "86": "PUNCT",
-    "87": "SCONJ",
-    "88": "SYM",
-    "89": "VERB",
-    "90": "VERB+AUX",
-    "91": "VERB+AUX+PART",
-    "92": "X"
+    "87": "PUNCT+PUNCT",
+    "88": "PUNCT+PUNCT+PUNCT",
+    "89": "SCONJ",
+    "90": "SYM",
+    "91": "VERB",
+    "92": "VERB+AUX",
+    "93": "VERB+AUX+PART",
+    "94": "X"
   },
   "initializer_range": 0.02,
   "intermediate_size": 768,
@@ -137,7 +139,7 @@
     "B-PART": 28,
     "B-PRON": 29,
     "B-PROPN": 30,
-    "B-PUNCT": 31,
+    "B-PUNCT+NOUN": 31,
     "B-SCONJ": 32,
     "B-SYM": 33,
     "B-VERB": 34,
@@ -174,7 +176,7 @@
     "I-PART": 65,
     "I-PRON": 66,
     "I-PROPN": 67,
-    "I-PUNCT": 68,
+    "I-PUNCT+NOUN": 68,
     "I-SCONJ": 69,
     "I-SYM": 70,
     "I-VERB": 71,
@@ -193,12 +195,14 @@
     "PRON+ADP": 84,
     "PROPN": 85,
     "PUNCT": 86,
-    "SCONJ": 87,
-    "SYM": 88,
-    "VERB": 89,
-    "VERB+AUX": 90,
-    "VERB+AUX+PART": 91,
-    "X": 92
+    "PUNCT+PUNCT": 87,
+    "PUNCT+PUNCT+PUNCT": 88,
+    "SCONJ": 89,
+    "SYM": 90,
+    "VERB": 91,
+    "VERB+AUX": 92,
+    "VERB+AUX+PART": 93,
+    "X": 94
   },
   "layer_norm_eps": 1e-07,
   "max_position_embeddings": 128,
@@ -210,9 +214,12 @@
   "pooler_dropout": 0,
   "pooler_hidden_act": "gelu",
   "pooler_hidden_size": 256,
-  "pos_att_type": null,
-  "position_biased_input": true,
-  "relative_attention": false,
+  "pos_att_type": [
+    "p2c",
+    "c2p"
+  ],
+  "position_biased_input": false,
+  "relative_attention": true,
   "task_specific_params": {
     "upos_multiword": {
       "ADJ+ADJ": {
@@ -364,6 +371,18 @@
           "\u304b"
         ]
       },
+      "PUNCT+NOUN": {
+        ",\u201c\u62c9\u81f4\u76e3\u7981\u201d\u554f\u984c": [
+          ",",
+          "\u201c\u62c9\u81f4\u76e3\u7981\u201d\u554f\u984c"
+        ]
+      },
+      "PUNCT+PUNCT": {
+        ",\u201c": [
+          ",",
+          "\u201c"
+        ]
+      },
       "VERB+AUX": {
         "\u3044\u308f\u308c": [
           "\u3044\u308f",
@@ -728,7 +747,7 @@
   },
   "tokenizer_class": "DebertaV2TokenizerFast",
   "torch_dtype": "float32",
-  "transformers_version": "4.19.1",
+  "transformers_version": "4.22.1",
   "type_vocab_size": 0,
   "vocab_size": 32000
 }

pytorch_model.bin

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9f130dada2cd229239d695632a67097b8469f20e42435e3e45f4f5a6cbf4435b
-size 64669363
+oid sha256:d92ef47bf3e9bf6c5a30f1bcb641a7e6f9753e12048d6316b730cdb93ea5b23a
+size 71135251

special_tokens_map.json

@@ -1 +1,9 @@
-{"bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]"}
+{
+  "bos_token": "[CLS]",
+  "cls_token": "[CLS]",
+  "eos_token": "[SEP]",
+  "mask_token": "[MASK]",
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "unk_token": "[UNK]"
+}

supar.model

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:2be76956d4705737f83345962fab32b8e99f0af5a0f3eb4a2615adb9ec1169f4
-size 113144939
+oid sha256:2b2bf72d53af75ee5f30f2be07e46b4abd4478d2166ff18199c96660105f5d9f
+size 119649355

tokenizer_config.json

@@ -1 +1,14 @@
-{"do_lower_case": false, "bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": "[MASK]", "split_by_punct": true, "keep_accents": true, "model_max_length": 128, "tokenizer_class": "DebertaV2TokenizerFast"}
+{
+  "bos_token": "[CLS]",
+  "cls_token": "[CLS]",
+  "do_lower_case": false,
+  "eos_token": "[SEP]",
+  "keep_accents": true,
+  "mask_token": "[MASK]",
+  "model_max_length": 128,
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "split_by_punct": true,
+  "tokenizer_class": "DebertaV2TokenizerFast",
+  "unk_token": "[UNK]"
+}