initial release

README.md

@@ -0,0 +1,33 @@
+---
+language:
+- "lzh"
+tags:
+- "classical chinese"
+- "literary chinese"
+- "ancient chinese"
+- "token-classification"
+- "pos"
+- "dependency-parsing"
+base_model: KoichiYasuoka/modernbert-base-classical-chinese
+datasets:
+- "universal_dependencies"
+license: "apache-2.0"
+pipeline_tag: "token-classification"
+widget:
+- text: "孟子見梁惠王"
+---
+
+# modernbert-base-classical-chinese-ud-embeds
+
+## Model Description
+
+This is a ModernBERT model pre-trained on Classical Chinese texts for POS-tagging and dependency-parsing, derived from [modernbert-base-classical-chinese](https://huggingface.co/KoichiYasuoka/modernbert-base-classical-chinese) and [UD_Classical_Chinese-Kyoto](https://github.com/UniversalDependencies/UD_Classical_Chinese-Kyoto).
+
+## How to Use
+
+```py
+from transformers import pipeline
+nlp=pipeline("universal-dependencies","KoichiYasuoka/modernbert-base-classical-chinese-ud-embeds",trust_remote_code=True)
+print(nlp("孟子見梁惠王"))
+```
+

config.json

@@ -0,0 +1,1643 @@
+{
+  "architectures": [
+    "ModernBertForTokenClassification"
+  ],
+  "attention_bias": false,
+  "attention_dropout": 0.0,
+  "bos_token_id": 0,
+  "classifier_activation": "gelu",
+  "classifier_bias": false,
+  "classifier_dropout": 0.0,
+  "classifier_pooling": "mean",
+  "cls_token_id": 0,
+  "custom_pipelines": {
+    "upos": {
+      "impl": "ud.BellmanFordTokenClassificationPipeline",
+      "pt": "AutoModelForTokenClassification"
+    },
+    "universal-dependencies": {
+      "impl": "ud.UniversalDependenciesPipeline",
+      "pt": "AutoModelForTokenClassification"
+    }
+  },
+  "decoder_bias": true,
+  "deterministic_flash_attn": false,
+  "embedding_dropout": 0.0,
+  "eos_token_id": 2,
+  "global_attn_every_n_layers": 3,
+  "global_rope_theta": 160000.0,
+  "gradient_checkpointing": false,
+  "hidden_activation": "gelu",
+  "hidden_size": 768,
+  "id2label": {
+    "0": "ADP",
+    "1": "ADP.",
+    "2": "ADP|Degree=Equ|_",
+    "3": "ADP|Degree=Equ|l-cc",
+    "4": "ADP|_",
+    "5": "ADP|l-acl",
+    "6": "ADP|l-advcl",
+    "7": "ADP|l-amod",
+    "8": "ADP|l-case",
+    "9": "ADP|l-cc",
+    "10": "ADP|l-mark",
+    "11": "ADP|l-nsubj",
+    "12": "ADP|l-obl",
+    "13": "ADP|r-case",
+    "14": "ADP|r-conj",
+    "15": "ADP|r-fixed",
+    "16": "ADP|r-mark",
+    "17": "ADP|r-obj",
+    "18": "ADP|root",
+    "19": "ADV",
+    "20": "ADV.",
+    "21": "ADV|AdvType=Cau|_",
+    "22": "ADV|AdvType=Cau|l-advmod",
+    "23": "ADV|AdvType=Cau|l-amod",
+    "24": "ADV|AdvType=Cau|l-nsubj",
+    "25": "ADV|AdvType=Cau|l-obj",
+    "26": "ADV|AdvType=Deg|Degree=Cmp|_",
+    "27": "ADV|AdvType=Deg|Degree=Cmp|l-advmod",
+    "28": "ADV|AdvType=Deg|Degree=Cmp|l-amod",
+    "29": "ADV|AdvType=Deg|Degree=Cmp|r-conj",
+    "30": "ADV|AdvType=Deg|Degree=Cmp|r-obj",
+    "31": "ADV|AdvType=Deg|Degree=Pos|_",
+    "32": "ADV|AdvType=Deg|Degree=Pos|l-advmod",
+    "33": "ADV|AdvType=Deg|Degree=Pos|l-amod",
+    "34": "ADV|AdvType=Deg|Degree=Pos|r-ccomp",
+    "35": "ADV|AdvType=Deg|Degree=Pos|r-conj",
+    "36": "ADV|AdvType=Deg|Degree=Pos|r-flat:vv",
+    "37": "ADV|AdvType=Deg|Degree=Pos|r-parataxis",
+    "38": "ADV|AdvType=Deg|Degree=Pos|root",
+    "39": "ADV|AdvType=Deg|Degree=Sup|_",
+    "40": "ADV|AdvType=Deg|Degree=Sup|l-advmod",
+    "41": "ADV|AdvType=Deg|Degree=Sup|l-amod",
+    "42": "ADV|AdvType=Deg|Degree=Sup|l-nsubj",
+    "43": "ADV|AdvType=Deg|Degree=Sup|r-conj",
+    "44": "ADV|AdvType=Deg|Degree=Sup|r-parataxis",
+    "45": "ADV|AdvType=Deg|Degree=Sup|root",
+    "46": "ADV|AdvType=Tim|Aspect=Perf|_",
+    "47": "ADV|AdvType=Tim|Aspect=Perf|l-advmod",
+    "48": "ADV|AdvType=Tim|Aspect=Perf|l-amod",
+    "49": "ADV|AdvType=Tim|Aspect=Perf|l-obl:lmod",
+    "50": "ADV|AdvType=Tim|Aspect=Perf|r-parataxis",
+    "51": "ADV|AdvType=Tim|Aspect=Perf|root",
+    "52": "ADV|AdvType=Tim|Tense=Fut|_",
+    "53": "ADV|AdvType=Tim|Tense=Fut|l-advmod",
+    "54": "ADV|AdvType=Tim|Tense=Fut|l-amod",
+    "55": "ADV|AdvType=Tim|Tense=Fut|l-nsubj",
+    "56": "ADV|AdvType=Tim|Tense=Fut|l-nsubj:outer",
+    "57": "ADV|AdvType=Tim|Tense=Fut|root",
+    "58": "ADV|AdvType=Tim|Tense=Past|_",
+    "59": "ADV|AdvType=Tim|Tense=Past|l-advmod",
+    "60": "ADV|AdvType=Tim|Tense=Past|l-amod",
+    "61": "ADV|AdvType=Tim|Tense=Pres|_",
+    "62": "ADV|AdvType=Tim|Tense=Pres|l-advmod",
+    "63": "ADV|AdvType=Tim|Tense=Pres|l-amod",
+    "64": "ADV|AdvType=Tim|Tense=Pres|root",
+    "65": "ADV|AdvType=Tim|_",
+    "66": "ADV|AdvType=Tim|l-advcl",
+    "67": "ADV|AdvType=Tim|l-advmod",
+    "68": "ADV|AdvType=Tim|l-amod",
+    "69": "ADV|AdvType=Tim|l-nsubj",
+    "70": "ADV|AdvType=Tim|r-advmod",
+    "71": "ADV|AdvType=Tim|r-ccomp",
+    "72": "ADV|AdvType=Tim|r-compound:redup",
+    "73": "ADV|AdvType=Tim|r-conj",
+    "74": "ADV|AdvType=Tim|r-flat:vv",
+    "75": "ADV|AdvType=Tim|r-parataxis",
+    "76": "ADV|AdvType=Tim|root",
+    "77": "ADV|Degree=Equ|VerbForm=Conv|_",
+    "78": "ADV|Degree=Equ|VerbForm=Conv|l-advmod",
+    "79": "ADV|Degree=Pos|VerbForm=Conv|_",
+    "80": "ADV|Degree=Pos|VerbForm=Conv|l-advmod",
+    "81": "ADV|Degree=Pos|VerbForm=Conv|r-advmod",
+    "82": "ADV|Polarity=Neg|VerbForm=Conv|_",
+    "83": "ADV|Polarity=Neg|VerbForm=Conv|l-advmod",
+    "84": "ADV|Polarity=Neg|_",
+    "85": "ADV|Polarity=Neg|l-advmod",
+    "86": "ADV|Polarity=Neg|l-amod",
+    "87": "ADV|Polarity=Neg|l-nsubj",
+    "88": "ADV|Polarity=Neg|l-parataxis",
+    "89": "ADV|Polarity=Neg|r-advmod",
+    "90": "ADV|Polarity=Neg|r-conj",
+    "91": "ADV|Polarity=Neg|r-obj",
+    "92": "ADV|Polarity=Neg|r-parataxis",
+    "93": "ADV|Polarity=Neg|root",
+    "94": "ADV|VerbForm=Conv|_",
+    "95": "ADV|VerbForm=Conv|l-advmod",
+    "96": "ADV|VerbForm=Conv|r-advmod",
+    "97": "ADV|_",
+    "98": "ADV|l-acl",
+    "99": "ADV|l-advcl",
+    "100": "ADV|l-advmod",
+    "101": "ADV|l-amod",
+    "102": "ADV|l-cc",
+    "103": "ADV|l-nsubj",
+    "104": "ADV|r-advmod",
+    "105": "ADV|r-ccomp",
+    "106": "ADV|r-conj",
+    "107": "ADV|r-flat:vv",
+    "108": "ADV|r-obj",
+    "109": "ADV|root",
+    "110": "AUX",
+    "111": "AUX.",
+    "112": "AUX|Mood=Des|_",
+    "113": "AUX|Mood=Des|l-aux",
+    "114": "AUX|Mood=Des|l-csubj",
+    "115": "AUX|Mood=Des|l-parataxis",
+    "116": "AUX|Mood=Des|r-ccomp",
+    "117": "AUX|Mood=Des|r-conj",
+    "118": "AUX|Mood=Des|r-flat:vv",
+    "119": "AUX|Mood=Des|root",
+    "120": "AUX|Mood=Nec|_",
+    "121": "AUX|Mood=Nec|l-acl",
+    "122": "AUX|Mood=Nec|l-amod",
+    "123": "AUX|Mood=Nec|l-aux",
+    "124": "AUX|Mood=Nec|r-aux",
+    "125": "AUX|Mood=Nec|root",
+    "126": "AUX|Mood=Pot|_",
+    "127": "AUX|Mood=Pot|l-acl",
+    "128": "AUX|Mood=Pot|l-advcl",
+    "129": "AUX|Mood=Pot|l-amod",
+    "130": "AUX|Mood=Pot|l-aux",
+    "131": "AUX|Mood=Pot|l-csubj",
+    "132": "AUX|Mood=Pot|l-nsubj",
+    "133": "AUX|Mood=Pot|r-ccomp",
+    "134": "AUX|Mood=Pot|r-conj",
+    "135": "AUX|Mood=Pot|r-obj",
+    "136": "AUX|Mood=Pot|r-parataxis",
+    "137": "AUX|Mood=Pot|r-xcomp",
+    "138": "AUX|Mood=Pot|root",
+    "139": "AUX|VerbType=Cop|_",
+    "140": "AUX|VerbType=Cop|l-cop",
+    "141": "AUX|Voice=Pass|_",
+    "142": "AUX|Voice=Pass|l-aux",
+    "143": "AUX|Voice=Pass|r-conj",
+    "144": "AUX|Voice=Pass|root",
+    "145": "B-ADP",
+    "146": "B-ADP.",
+    "147": "B-ADV",
+    "148": "B-ADV.",
+    "149": "B-AUX",
+    "150": "B-AUX.",
+    "151": "B-CCONJ",
+    "152": "B-CCONJ.",
+    "153": "B-INTJ",
+    "154": "B-INTJ.",
+    "155": "B-NOUN",
+    "156": "B-NOUN.",
+    "157": "B-NUM",
+    "158": "B-NUM.",
+    "159": "B-PART",
+    "160": "B-PART.",
+    "161": "B-PRON",
+    "162": "B-PRON.",
+    "163": "B-PROPN",
+    "164": "B-PROPN.",
+    "165": "B-PUNCT",
+    "166": "B-PUNCT.",
+    "167": "B-SCONJ",
+    "168": "B-SCONJ.",
+    "169": "B-SYM",
+    "170": "B-SYM.",
+    "171": "B-VERB",
+    "172": "B-VERB.",
+    "173": "CCONJ",
+    "174": "CCONJ.",
+    "175": "CCONJ|_",
+    "176": "CCONJ|l-advmod",
+    "177": "CCONJ|l-amod",
+    "178": "CCONJ|l-cc",
+    "179": "CCONJ|l-obj",
+    "180": "CCONJ|r-fixed",
+    "181": "CCONJ|r-orphan",
+    "182": "I-ADP",
+    "183": "I-ADP.",
+    "184": "I-ADV",
+    "185": "I-ADV.",
+    "186": "I-AUX",
+    "187": "I-AUX.",
+    "188": "I-CCONJ",
+    "189": "I-CCONJ.",
+    "190": "I-INTJ",
+    "191": "I-INTJ.",
+    "192": "I-NOUN",
+    "193": "I-NOUN.",
+    "194": "I-NUM",
+    "195": "I-NUM.",
+    "196": "I-PART",
+    "197": "I-PART.",
+    "198": "I-PRON",
+    "199": "I-PRON.",
+    "200": "I-PROPN",
+    "201": "I-PROPN.",
+    "202": "I-PUNCT",
+    "203": "I-PUNCT.",
+    "204": "I-SCONJ",
+    "205": "I-SCONJ.",
+    "206": "I-SYM",
+    "207": "I-SYM.",
+    "208": "I-VERB",
+    "209": "I-VERB.",
+    "210": "INTJ",
+    "211": "INTJ.",
+    "212": "INTJ|_",
+    "213": "INTJ|l-advcl",
+    "214": "INTJ|l-csubj",
+    "215": "INTJ|l-discourse",
+    "216": "INTJ|l-discourse:sp",
+    "217": "INTJ|l-dislocated",
+    "218": "INTJ|l-nsubj",
+    "219": "INTJ|l-vocative",
+    "220": "INTJ|r-compound:redup",
+    "221": "INTJ|r-conj",
+    "222": "INTJ|r-discourse:sp",
+    "223": "INTJ|r-dislocated",
+    "224": "INTJ|r-fixed",
+    "225": "INTJ|r-obj",
+    "226": "INTJ|r-parataxis",
+    "227": "INTJ|root",
+    "228": "NOUN",
+    "229": "NOUN.",
+    "230": "NOUN|Case=Loc|_",
+    "231": "NOUN|Case=Loc|l-acl",
+    "232": "NOUN|Case=Loc|l-advcl",
+    "233": "NOUN|Case=Loc|l-amod",
+    "234": "NOUN|Case=Loc|l-clf",
+    "235": "NOUN|Case=Loc|l-compound",
+    "236": "NOUN|Case=Loc|l-csubj",
+    "237": "NOUN|Case=Loc|l-dislocated",
+    "238": "NOUN|Case=Loc|l-nmod",
+    "239": "NOUN|Case=Loc|l-nsubj",
+    "240": "NOUN|Case=Loc|l-nsubj:outer",
+    "241": "NOUN|Case=Loc|l-obj",
+    "242": "NOUN|Case=Loc|l-obl",
+    "243": "NOUN|Case=Loc|l-obl:lmod",
+    "244": "NOUN|Case=Loc|l-obl:tmod",
+    "245": "NOUN|Case=Loc|l-parataxis",
+    "246": "NOUN|Case=Loc|r-ccomp",
+    "247": "NOUN|Case=Loc|r-clf",
+    "248": "NOUN|Case=Loc|r-compound:redup",
+    "249": "NOUN|Case=Loc|r-conj",
+    "250": "NOUN|Case=Loc|r-dislocated",
+    "251": "NOUN|Case=Loc|r-flat",
+    "252": "NOUN|Case=Loc|r-iobj",
+    "253": "NOUN|Case=Loc|r-list",
+    "254": "NOUN|Case=Loc|r-nmod",
+    "255": "NOUN|Case=Loc|r-nsubj",
+    "256": "NOUN|Case=Loc|r-obj",
+    "257": "NOUN|Case=Loc|r-obl",
+    "258": "NOUN|Case=Loc|r-obl:lmod",
+    "259": "NOUN|Case=Loc|r-parataxis",
+    "260": "NOUN|Case=Loc|r-xcomp",
+    "261": "NOUN|Case=Loc|root",
+    "262": "NOUN|Case=Tem|_",
+    "263": "NOUN|Case=Tem|l-acl",
+    "264": "NOUN|Case=Tem|l-advcl",
+    "265": "NOUN|Case=Tem|l-amod",
+    "266": "NOUN|Case=Tem|l-compound",
+    "267": "NOUN|Case=Tem|l-csubj",
+    "268": "NOUN|Case=Tem|l-nmod",
+    "269": "NOUN|Case=Tem|l-nsubj",
+    "270": "NOUN|Case=Tem|l-nsubj:outer",
+    "271": "NOUN|Case=Tem|l-obj",
+    "272": "NOUN|Case=Tem|l-obl:tmod",
+    "273": "NOUN|Case=Tem|r-amod",
+    "274": "NOUN|Case=Tem|r-ccomp",
+    "275": "NOUN|Case=Tem|r-clf",
+    "276": "NOUN|Case=Tem|r-compound:redup",
+    "277": "NOUN|Case=Tem|r-conj",
+    "278": "NOUN|Case=Tem|r-flat",
+    "279": "NOUN|Case=Tem|r-iobj",
+    "280": "NOUN|Case=Tem|r-list",
+    "281": "NOUN|Case=Tem|r-nsubj",
+    "282": "NOUN|Case=Tem|r-obj",
+    "283": "NOUN|Case=Tem|r-obl:tmod",
+    "284": "NOUN|Case=Tem|r-parataxis",
+    "285": "NOUN|Case=Tem|r-xcomp",
+    "286": "NOUN|Case=Tem|root",
+    "287": "NOUN|Degree=Pos|_",
+    "288": "NOUN|Degree=Pos|root",
+    "289": "NOUN|NounType=Clf|_",
+    "290": "NOUN|NounType=Clf|l-clf",
+    "291": "NOUN|NounType=Clf|l-nmod",
+    "292": "NOUN|NounType=Clf|l-nsubj",
+    "293": "NOUN|NounType=Clf|l-obl",
+    "294": "NOUN|NounType=Clf|r-ccomp",
+    "295": "NOUN|NounType=Clf|r-clf",
+    "296": "NOUN|NounType=Clf|r-compound:redup",
+    "297": "NOUN|NounType=Clf|r-conj",
+    "298": "NOUN|NounType=Clf|r-flat",
+    "299": "NOUN|NounType=Clf|r-obj",
+    "300": "NOUN|NounType=Clf|r-parataxis",
+    "301": "NOUN|NounType=Clf|root",
+    "302": "NOUN|_",
+    "303": "NOUN|l-acl",
+    "304": "NOUN|l-advcl",
+    "305": "NOUN|l-amod",
+    "306": "NOUN|l-ccomp",
+    "307": "NOUN|l-clf",
+    "308": "NOUN|l-compound",
+    "309": "NOUN|l-csubj",
+    "310": "NOUN|l-csubj:outer",
+    "311": "NOUN|l-dislocated",
+    "312": "NOUN|l-iobj",
+    "313": "NOUN|l-list",
+    "314": "NOUN|l-nmod",
+    "315": "NOUN|l-nsubj",
+    "316": "NOUN|l-nsubj:outer",
+    "317": "NOUN|l-nsubj:pass",
+    "318": "NOUN|l-obj",
+    "319": "NOUN|l-obl",
+    "320": "NOUN|l-obl:lmod",
+    "321": "NOUN|l-obl:tmod",
+    "322": "NOUN|l-vocative",
+    "323": "NOUN|r-acl",
+    "324": "NOUN|r-advcl",
+    "325": "NOUN|r-amod",
+    "326": "NOUN|r-ccomp",
+    "327": "NOUN|r-clf",
+    "328": "NOUN|r-compound:redup",
+    "329": "NOUN|r-conj",
+    "330": "NOUN|r-csubj",
+    "331": "NOUN|r-dislocated",
+    "332": "NOUN|r-flat",
+    "333": "NOUN|r-flat:foreign",
+    "334": "NOUN|r-iobj",
+    "335": "NOUN|r-list",
+    "336": "NOUN|r-nmod",
+    "337": "NOUN|r-nsubj",
+    "338": "NOUN|r-obj",
+    "339": "NOUN|r-obl",
+    "340": "NOUN|r-obl:lmod",
+    "341": "NOUN|r-parataxis",
+    "342": "NOUN|r-vocative",
+    "343": "NOUN|r-xcomp",
+    "344": "NOUN|root",
+    "345": "NUM",
+    "346": "NUM.",
+    "347": "NUM|NumType=Ord|_",
+    "348": "NUM|NumType=Ord|l-nsubj",
+    "349": "NUM|NumType=Ord|l-nummod",
+    "350": "NUM|NumType=Ord|l-obl",
+    "351": "NUM|NumType=Ord|l-obl:lmod",
+    "352": "NUM|NumType=Ord|l-obl:tmod",
+    "353": "NUM|NumType=Ord|r-conj",
+    "354": "NUM|NumType=Ord|r-flat",
+    "355": "NUM|NumType=Ord|r-obj",
+    "356": "NUM|NumType=Ord|root",
+    "357": "NUM|_",
+    "358": "NUM|l-acl",
+    "359": "NUM|l-advcl",
+    "360": "NUM|l-compound",
+    "361": "NUM|l-csubj",
+    "362": "NUM|l-dislocated",
+    "363": "NUM|l-nsubj",
+    "364": "NUM|l-nsubj:outer",
+    "365": "NUM|l-nummod",
+    "366": "NUM|l-obj",
+    "367": "NUM|l-obl",
+    "368": "NUM|l-obl:lmod",
+    "369": "NUM|l-obl:tmod",
+    "370": "NUM|r-ccomp",
+    "371": "NUM|r-clf",
+    "372": "NUM|r-compound",
+    "373": "NUM|r-compound:redup",
+    "374": "NUM|r-conj",
+    "375": "NUM|r-flat",
+    "376": "NUM|r-iobj",
+    "377": "NUM|r-list",
+    "378": "NUM|r-nummod",
+    "379": "NUM|r-obj",
+    "380": "NUM|r-obl",
+    "381": "NUM|r-obl:tmod",
+    "382": "NUM|r-parataxis",
+    "383": "NUM|r-xcomp",
+    "384": "NUM|root",
+    "385": "PART",
+    "386": "PART.",
+    "387": "PART|_",
+    "388": "PART|l-acl",
+    "389": "PART|l-advcl",
+    "390": "PART|l-advmod",
+    "391": "PART|l-amod",
+    "392": "PART|l-case",
+    "393": "PART|l-cc",
+    "394": "PART|l-csubj",
+    "395": "PART|l-csubj:outer",
+    "396": "PART|l-discourse",
+    "397": "PART|l-discourse:sp",
+    "398": "PART|l-dislocated",
+    "399": "PART|l-mark",
+    "400": "PART|l-nmod",
+    "401": "PART|l-nsubj",
+    "402": "PART|l-nsubj:outer",
+    "403": "PART|l-nsubj:pass",
+    "404": "PART|l-obj",
+    "405": "PART|l-obl",
+    "406": "PART|l-obl:lmod",
+    "407": "PART|r-advmod",
+    "408": "PART|r-case",
+    "409": "PART|r-ccomp",
+    "410": "PART|r-clf",
+    "411": "PART|r-conj",
+    "412": "PART|r-discourse",
+    "413": "PART|r-discourse:sp",
+    "414": "PART|r-dislocated",
+    "415": "PART|r-fixed",
+    "416": "PART|r-flat",
+    "417": "PART|r-iobj",
+    "418": "PART|r-list",
+    "419": "PART|r-mark",
+    "420": "PART|r-nsubj",
+    "421": "PART|r-obj",
+    "422": "PART|r-obl",
+    "423": "PART|r-parataxis",
+    "424": "PART|r-xcomp",
+    "425": "PART|root",
+    "426": "PRON",
+    "427": "PRON.",
+    "428": "PRON|Person=1|PronType=Prs|_",
+    "429": "PRON|Person=1|PronType=Prs|l-acl",
+    "430": "PRON|Person=1|PronType=Prs|l-advcl",
+    "431": "PRON|Person=1|PronType=Prs|l-det",
+    "432": "PRON|Person=1|PronType=Prs|l-iobj",
+    "433": "PRON|Person=1|PronType=Prs|l-nsubj",
+    "434": "PRON|Person=1|PronType=Prs|l-nsubj:outer",
+    "435": "PRON|Person=1|PronType=Prs|l-obj",
+    "436": "PRON|Person=1|PronType=Prs|l-obl",
+    "437": "PRON|Person=1|PronType=Prs|l-vocative",
+    "438": "PRON|Person=1|PronType=Prs|r-ccomp",
+    "439": "PRON|Person=1|PronType=Prs|r-conj",
+    "440": "PRON|Person=1|PronType=Prs|r-iobj",
+    "441": "PRON|Person=1|PronType=Prs|r-nsubj",
+    "442": "PRON|Person=1|PronType=Prs|r-obj",
+    "443": "PRON|Person=1|PronType=Prs|r-obl",
+    "444": "PRON|Person=1|PronType=Prs|r-obl:lmod",
+    "445": "PRON|Person=1|PronType=Prs|root",
+    "446": "PRON|Person=2|PronType=Prs|_",
+    "447": "PRON|Person=2|PronType=Prs|l-advcl",
+    "448": "PRON|Person=2|PronType=Prs|l-amod",
+    "449": "PRON|Person=2|PronType=Prs|l-det",
+    "450": "PRON|Person=2|PronType=Prs|l-nmod",
+    "451": "PRON|Person=2|PronType=Prs|l-nsubj",
+    "452": "PRON|Person=2|PronType=Prs|l-nsubj:outer",
+    "453": "PRON|Person=2|PronType=Prs|l-obj",
+    "454": "PRON|Person=2|PronType=Prs|l-obl",
+    "455": "PRON|Person=2|PronType=Prs|l-vocative",
+    "456": "PRON|Person=2|PronType=Prs|r-conj",
+    "457": "PRON|Person=2|PronType=Prs|r-flat",
+    "458": "PRON|Person=2|PronType=Prs|r-iobj",
+    "459": "PRON|Person=2|PronType=Prs|r-obj",
+    "460": "PRON|Person=2|PronType=Prs|r-obl",
+    "461": "PRON|Person=2|PronType=Prs|root",
+    "462": "PRON|Person=3|PronType=Prs|_",
+    "463": "PRON|Person=3|PronType=Prs|l-advcl",
+    "464": "PRON|Person=3|PronType=Prs|l-amod",
+    "465": "PRON|Person=3|PronType=Prs|l-det",
+    "466": "PRON|Person=3|PronType=Prs|l-dislocated",
+    "467": "PRON|Person=3|PronType=Prs|l-expl",
+    "468": "PRON|Person=3|PronType=Prs|l-iobj",
+    "469": "PRON|Person=3|PronType=Prs|l-nsubj",
+    "470": "PRON|Person=3|PronType=Prs|l-nsubj:outer",
+    "471": "PRON|Person=3|PronType=Prs|l-nsubj:pass",
+    "472": "PRON|Person=3|PronType=Prs|l-obj",
+    "473": "PRON|Person=3|PronType=Prs|l-obl",
+    "474": "PRON|Person=3|PronType=Prs|r-ccomp",
+    "475": "PRON|Person=3|PronType=Prs|r-conj",
+    "476": "PRON|Person=3|PronType=Prs|r-expl",
+    "477": "PRON|Person=3|PronType=Prs|r-iobj",
+    "478": "PRON|Person=3|PronType=Prs|r-nsubj",
+    "479": "PRON|Person=3|PronType=Prs|r-obj",
+    "480": "PRON|Person=3|PronType=Prs|r-obl",
+    "481": "PRON|Person=3|PronType=Prs|root",
+    "482": "PRON|PronType=Dem|_",
+    "483": "PRON|PronType=Dem|l-acl",
+    "484": "PRON|PronType=Dem|l-advcl",
+    "485": "PRON|PronType=Dem|l-amod",
+    "486": "PRON|PronType=Dem|l-compound",
+    "487": "PRON|PronType=Dem|l-det",
+    "488": "PRON|PronType=Dem|l-dislocated",
+    "489": "PRON|PronType=Dem|l-expl",
+    "490": "PRON|PronType=Dem|l-nsubj",
+    "491": "PRON|PronType=Dem|l-nsubj:outer",
+    "492": "PRON|PronType=Dem|l-obj",
+    "493": "PRON|PronType=Dem|l-obl",
+    "494": "PRON|PronType=Dem|l-obl:lmod",
+    "495": "PRON|PronType=Dem|r-conj",
+    "496": "PRON|PronType=Dem|r-det",
+    "497": "PRON|PronType=Dem|r-expl",
+    "498": "PRON|PronType=Dem|r-flat",
+    "499": "PRON|PronType=Dem|r-iobj",
+    "500": "PRON|PronType=Dem|r-obj",
+    "501": "PRON|PronType=Dem|r-obl",
+    "502": "PRON|PronType=Dem|r-obl:lmod",
+    "503": "PRON|PronType=Dem|root",
+    "504": "PRON|PronType=Int|_",
+    "505": "PRON|PronType=Int|l-advcl",
+    "506": "PRON|PronType=Int|l-amod",
+    "507": "PRON|PronType=Int|l-det",
+    "508": "PRON|PronType=Int|l-dislocated",
+    "509": "PRON|PronType=Int|l-nsubj",
+    "510": "PRON|PronType=Int|l-nsubj:outer",
+    "511": "PRON|PronType=Int|l-obj",
+    "512": "PRON|PronType=Int|l-obl",
+    "513": "PRON|PronType=Int|l-vocative",
+    "514": "PRON|PronType=Int|r-ccomp",
+    "515": "PRON|PronType=Int|r-conj",
+    "516": "PRON|PronType=Int|r-flat",
+    "517": "PRON|PronType=Int|r-obj",
+    "518": "PRON|PronType=Int|r-parataxis",
+    "519": "PRON|PronType=Int|r-xcomp",
+    "520": "PRON|PronType=Int|root",
+    "521": "PRON|PronType=Prs|Reflex=Yes|_",
+    "522": "PRON|PronType=Prs|Reflex=Yes|l-acl",
+    "523": "PRON|PronType=Prs|Reflex=Yes|l-det",
+    "524": "PRON|PronType=Prs|Reflex=Yes|l-nsubj",
+    "525": "PRON|PronType=Prs|Reflex=Yes|l-obj",
+    "526": "PRON|PronType=Prs|Reflex=Yes|l-obl",
+    "527": "PRON|PronType=Prs|Reflex=Yes|r-dislocated",
+    "528": "PRON|PronType=Prs|Reflex=Yes|r-obj",
+    "529": "PRON|PronType=Prs|Reflex=Yes|r-obl",
+    "530": "PRON|PronType=Prs|Reflex=Yes|root",
+    "531": "PRON|PronType=Prs|_",
+    "532": "PRON|PronType=Prs|l-det",
+    "533": "PRON|PronType=Prs|l-nsubj",
+    "534": "PRON|PronType=Prs|l-nsubj:outer",
+    "535": "PRON|PronType=Prs|l-obj",
+    "536": "PRON|PronType=Prs|r-conj",
+    "537": "PRON|PronType=Prs|r-iobj",
+    "538": "PRON|PronType=Prs|r-obj",
+    "539": "PROPN",
+    "540": "PROPN.",
+    "541": "PROPN|Case=Loc|NameType=Geo|_",
+    "542": "PROPN|Case=Loc|NameType=Geo|l-acl",
+    "543": "PROPN|Case=Loc|NameType=Geo|l-advcl",
+    "544": "PROPN|Case=Loc|NameType=Geo|l-amod",
+    "545": "PROPN|Case=Loc|NameType=Geo|l-compound",
+    "546": "PROPN|Case=Loc|NameType=Geo|l-csubj",
+    "547": "PROPN|Case=Loc|NameType=Geo|l-dislocated",
+    "548": "PROPN|Case=Loc|NameType=Geo|l-nmod",
+    "549": "PROPN|Case=Loc|NameType=Geo|l-nsubj",
+    "550": "PROPN|Case=Loc|NameType=Geo|l-nsubj:outer",
+    "551": "PROPN|Case=Loc|NameType=Geo|l-obl",
+    "552": "PROPN|Case=Loc|NameType=Geo|l-obl:lmod",
+    "553": "PROPN|Case=Loc|NameType=Geo|r-conj",
+    "554": "PROPN|Case=Loc|NameType=Geo|r-flat",
+    "555": "PROPN|Case=Loc|NameType=Geo|r-iobj",
+    "556": "PROPN|Case=Loc|NameType=Geo|r-obj",
+    "557": "PROPN|Case=Loc|NameType=Geo|r-obl",
+    "558": "PROPN|Case=Loc|NameType=Geo|r-obl:lmod",
+    "559": "PROPN|Case=Loc|NameType=Geo|r-parataxis",
+    "560": "PROPN|Case=Loc|NameType=Geo|r-xcomp",
+    "561": "PROPN|Case=Loc|NameType=Geo|root",
+    "562": "PROPN|Case=Loc|NameType=Nat|_",
+    "563": "PROPN|Case=Loc|NameType=Nat|l-acl",
+    "564": "PROPN|Case=Loc|NameType=Nat|l-advcl",
+    "565": "PROPN|Case=Loc|NameType=Nat|l-amod",
+    "566": "PROPN|Case=Loc|NameType=Nat|l-clf",
+    "567": "PROPN|Case=Loc|NameType=Nat|l-compound",
+    "568": "PROPN|Case=Loc|NameType=Nat|l-nmod",
+    "569": "PROPN|Case=Loc|NameType=Nat|l-nsubj",
+    "570": "PROPN|Case=Loc|NameType=Nat|l-nsubj:outer",
+    "571": "PROPN|Case=Loc|NameType=Nat|l-nsubj:pass",
+    "572": "PROPN|Case=Loc|NameType=Nat|l-obj",
+    "573": "PROPN|Case=Loc|NameType=Nat|l-obl",
+    "574": "PROPN|Case=Loc|NameType=Nat|l-obl:lmod",
+    "575": "PROPN|Case=Loc|NameType=Nat|r-ccomp",
+    "576": "PROPN|Case=Loc|NameType=Nat|r-conj",
+    "577": "PROPN|Case=Loc|NameType=Nat|r-flat",
+    "578": "PROPN|Case=Loc|NameType=Nat|r-iobj",
+    "579": "PROPN|Case=Loc|NameType=Nat|r-nmod",
+    "580": "PROPN|Case=Loc|NameType=Nat|r-obj",
+    "581": "PROPN|Case=Loc|NameType=Nat|r-obl",
+    "582": "PROPN|Case=Loc|NameType=Nat|r-obl:lmod",
+    "583": "PROPN|Case=Loc|NameType=Nat|r-parataxis",
+    "584": "PROPN|Case=Loc|NameType=Nat|r-xcomp",
+    "585": "PROPN|Case=Loc|NameType=Nat|root",
+    "586": "PROPN|NameType=Giv|_",
+    "587": "PROPN|NameType=Giv|l-acl",
+    "588": "PROPN|NameType=Giv|l-advcl",
+    "589": "PROPN|NameType=Giv|l-amod",
+    "590": "PROPN|NameType=Giv|l-compound",
+    "591": "PROPN|NameType=Giv|l-dislocated",
+    "592": "PROPN|NameType=Giv|l-nmod",
+    "593": "PROPN|NameType=Giv|l-nsubj",
+    "594": "PROPN|NameType=Giv|l-nsubj:outer",
+    "595": "PROPN|NameType=Giv|l-nsubj:pass",
+    "596": "PROPN|NameType=Giv|l-obj",
+    "597": "PROPN|NameType=Giv|l-obl",
+    "598": "PROPN|NameType=Giv|l-obl:lmod",
+    "599": "PROPN|NameType=Giv|l-parataxis",
+    "600": "PROPN|NameType=Giv|l-vocative",
+    "601": "PROPN|NameType=Giv|r-appos",
+    "602": "PROPN|NameType=Giv|r-ccomp",
+    "603": "PROPN|NameType=Giv|r-conj",
+    "604": "PROPN|NameType=Giv|r-dislocated",
+    "605": "PROPN|NameType=Giv|r-flat",
+    "606": "PROPN|NameType=Giv|r-iobj",
+    "607": "PROPN|NameType=Giv|r-list",
+    "608": "PROPN|NameType=Giv|r-nmod",
+    "609": "PROPN|NameType=Giv|r-obj",
+    "610": "PROPN|NameType=Giv|r-obl",
+    "611": "PROPN|NameType=Giv|r-obl:lmod",
+    "612": "PROPN|NameType=Giv|r-parataxis",
+    "613": "PROPN|NameType=Giv|r-xcomp",
+    "614": "PROPN|NameType=Giv|root",
+    "615": "PROPN|NameType=Prs|_",
+    "616": "PROPN|NameType=Prs|l-acl",
+    "617": "PROPN|NameType=Prs|l-advcl",
+    "618": "PROPN|NameType=Prs|l-amod",
+    "619": "PROPN|NameType=Prs|l-compound",
+    "620": "PROPN|NameType=Prs|l-dislocated",
+    "621": "PROPN|NameType=Prs|l-nmod",
+    "622": "PROPN|NameType=Prs|l-nsubj",
+    "623": "PROPN|NameType=Prs|l-nsubj:outer",
+    "624": "PROPN|NameType=Prs|l-obj",
+    "625": "PROPN|NameType=Prs|l-obl",
+    "626": "PROPN|NameType=Prs|r-conj",
+    "627": "PROPN|NameType=Prs|r-dislocated",
+    "628": "PROPN|NameType=Prs|r-flat",
+    "629": "PROPN|NameType=Prs|r-iobj",
+    "630": "PROPN|NameType=Prs|r-obj",
+    "631": "PROPN|NameType=Prs|r-obl",
+    "632": "PROPN|NameType=Prs|r-parataxis",
+    "633": "PROPN|NameType=Prs|root",
+    "634": "PROPN|NameType=Sur|_",
+    "635": "PROPN|NameType=Sur|l-acl",
+    "636": "PROPN|NameType=Sur|l-advcl",
+    "637": "PROPN|NameType=Sur|l-amod",
+    "638": "PROPN|NameType=Sur|l-compound",
+    "639": "PROPN|NameType=Sur|l-csubj",
+    "640": "PROPN|NameType=Sur|l-dislocated",
+    "641": "PROPN|NameType=Sur|l-nmod",
+    "642": "PROPN|NameType=Sur|l-nsubj",
+    "643": "PROPN|NameType=Sur|l-nsubj:outer",
+    "644": "PROPN|NameType=Sur|l-nsubj:pass",
+    "645": "PROPN|NameType=Sur|l-obl",
+    "646": "PROPN|NameType=Sur|l-obl:lmod",
+    "647": "PROPN|NameType=Sur|l-vocative",
+    "648": "PROPN|NameType=Sur|r-ccomp",
+    "649": "PROPN|NameType=Sur|r-conj",
+    "650": "PROPN|NameType=Sur|r-dislocated",
+    "651": "PROPN|NameType=Sur|r-flat",
+    "652": "PROPN|NameType=Sur|r-iobj",
+    "653": "PROPN|NameType=Sur|r-list",
+    "654": "PROPN|NameType=Sur|r-nmod",
+    "655": "PROPN|NameType=Sur|r-nsubj",
+    "656": "PROPN|NameType=Sur|r-obj",
+    "657": "PROPN|NameType=Sur|r-obl",
+    "658": "PROPN|NameType=Sur|r-obl:lmod",
+    "659": "PROPN|NameType=Sur|r-parataxis",
+    "660": "PROPN|NameType=Sur|r-xcomp",
+    "661": "PROPN|NameType=Sur|root",
+    "662": "PROPN|_",
+    "663": "PROPN|l-nmod",
+    "664": "PUNCT",
+    "665": "PUNCT.",
+    "666": "PUNCT|_",
+    "667": "PUNCT|root",
+    "668": "SCONJ",
+    "669": "SCONJ.",
+    "670": "SCONJ|_",
+    "671": "SCONJ|l-case",
+    "672": "SCONJ|l-cc",
+    "673": "SCONJ|l-mark",
+    "674": "SCONJ|l-nsubj",
+    "675": "SCONJ|l-obl",
+    "676": "SCONJ|r-case",
+    "677": "SCONJ|r-iobj",
+    "678": "SCONJ|r-mark",
+    "679": "SCONJ|r-nsubj",
+    "680": "SCONJ|r-nsubj:pass",
+    "681": "SCONJ|r-obj",
+    "682": "SCONJ|root",
+    "683": "SYM",
+    "684": "SYM.",
+    "685": "SYM|_",
+    "686": "SYM|l-nmod",
+    "687": "SYM|l-nsubj",
+    "688": "SYM|r-conj",
+    "689": "SYM|r-nmod",
+    "690": "SYM|r-xcomp",
+    "691": "SYM|root",
+    "692": "VERB",
+    "693": "VERB.",
+    "694": "VERB|Degree=Equ|VerbForm=Part|_",
+    "695": "VERB|Degree=Equ|VerbForm=Part|l-amod",
+    "696": "VERB|Degree=Equ|_",
+    "697": "VERB|Degree=Equ|l-acl",
+    "698": "VERB|Degree=Equ|l-advcl",
+    "699": "VERB|Degree=Equ|l-ccomp",
+    "700": "VERB|Degree=Equ|l-csubj",
+    "701": "VERB|Degree=Equ|l-nsubj",
+    "702": "VERB|Degree=Equ|l-obj",
+    "703": "VERB|Degree=Equ|r-ccomp",
+    "704": "VERB|Degree=Equ|r-compound:redup",
+    "705": "VERB|Degree=Equ|r-conj",
+    "706": "VERB|Degree=Equ|r-obj",
+    "707": "VERB|Degree=Equ|r-parataxis",
+    "708": "VERB|Degree=Equ|r-xcomp",
+    "709": "VERB|Degree=Equ|root",
+    "710": "VERB|Degree=Pos|VerbForm=Part|_",
+    "711": "VERB|Degree=Pos|VerbForm=Part|l-amod",
+    "712": "VERB|Degree=Pos|VerbForm=Part|r-amod",
+    "713": "VERB|Degree=Pos|_",
+    "714": "VERB|Degree=Pos|l-acl",
+    "715": "VERB|Degree=Pos|l-advcl",
+    "716": "VERB|Degree=Pos|l-ccomp",
+    "717": "VERB|Degree=Pos|l-csubj",
+    "718": "VERB|Degree=Pos|l-csubj:outer",
+    "719": "VERB|Degree=Pos|l-dislocated",
+    "720": "VERB|Degree=Pos|l-nsubj",
+    "721": "VERB|Degree=Pos|l-nsubj:outer",
+    "722": "VERB|Degree=Pos|l-obj",
+    "723": "VERB|Degree=Pos|l-obl",
+    "724": "VERB|Degree=Pos|l-vocative",
+    "725": "VERB|Degree=Pos|r-advcl",
+    "726": "VERB|Degree=Pos|r-ccomp",
+    "727": "VERB|Degree=Pos|r-compound:redup",
+    "728": "VERB|Degree=Pos|r-conj",
+    "729": "VERB|Degree=Pos|r-dislocated",
+    "730": "VERB|Degree=Pos|r-fixed",
+    "731": "VERB|Degree=Pos|r-flat:vv",
+    "732": "VERB|Degree=Pos|r-iobj",
+    "733": "VERB|Degree=Pos|r-obj",
+    "734": "VERB|Degree=Pos|r-obl",
+    "735": "VERB|Degree=Pos|r-parataxis",
+    "736": "VERB|Degree=Pos|r-xcomp",
+    "737": "VERB|Degree=Pos|root",
+    "738": "VERB|Polarity=Neg|VerbForm=Part|_",
+    "739": "VERB|Polarity=Neg|VerbForm=Part|l-amod",
+    "740": "VERB|Polarity=Neg|_",
+    "741": "VERB|Polarity=Neg|l-acl",
+    "742": "VERB|Polarity=Neg|l-advcl",
+    "743": "VERB|Polarity=Neg|l-ccomp",
+    "744": "VERB|Polarity=Neg|l-csubj",
+    "745": "VERB|Polarity=Neg|l-csubj:outer",
+    "746": "VERB|Polarity=Neg|l-nsubj",
+    "747": "VERB|Polarity=Neg|l-obl",
+    "748": "VERB|Polarity=Neg|r-advcl",
+    "749": "VERB|Polarity=Neg|r-ccomp",
+    "750": "VERB|Polarity=Neg|r-conj",
+    "751": "VERB|Polarity=Neg|r-flat:vv",
+    "752": "VERB|Polarity=Neg|r-obj",
+    "753": "VERB|Polarity=Neg|r-obl",
+    "754": "VERB|Polarity=Neg|r-parataxis",
+    "755": "VERB|Polarity=Neg|r-xcomp",
+    "756": "VERB|Polarity=Neg|root",
+    "757": "VERB|VerbForm=Part|_",
+    "758": "VERB|VerbForm=Part|l-amod",
+    "759": "VERB|VerbForm=Part|r-amod",
+    "760": "VERB|_",
+    "761": "VERB|l-acl",
+    "762": "VERB|l-advcl",
+    "763": "VERB|l-ccomp",
+    "764": "VERB|l-csubj",
+    "765": "VERB|l-csubj:outer",
+    "766": "VERB|l-csubj:pass",
+    "767": "VERB|l-dislocated",
+    "768": "VERB|l-nsubj",
+    "769": "VERB|l-nsubj:outer",
+    "770": "VERB|l-obj",
+    "771": "VERB|l-obl",
+    "772": "VERB|l-obl:lmod",
+    "773": "VERB|l-parataxis",
+    "774": "VERB|r-acl",
+    "775": "VERB|r-advcl",
+    "776": "VERB|r-ccomp",
+    "777": "VERB|r-compound:redup",
+    "778": "VERB|r-conj",
+    "779": "VERB|r-dislocated",
+    "780": "VERB|r-fixed",
+    "781": "VERB|r-flat:vv",
+    "782": "VERB|r-iobj",
+    "783": "VERB|r-list",
+    "784": "VERB|r-obj",
+    "785": "VERB|r-obl",
+    "786": "VERB|r-obl:lmod",
+    "787": "VERB|r-parataxis",
+    "788": "VERB|r-vocative",
+    "789": "VERB|r-xcomp",
+    "790": "VERB|root"
+  },
+  "initializer_cutoff_factor": 2.0,
+  "initializer_range": 0.02,
+  "intermediate_size": 1152,
+  "label2id": {
+    "ADP": 0,
+    "ADP.": 1,
+    "ADP|Degree=Equ|_": 2,
+    "ADP|Degree=Equ|l-cc": 3,
+    "ADP|_": 4,
+    "ADP|l-acl": 5,
+    "ADP|l-advcl": 6,
+    "ADP|l-amod": 7,
+    "ADP|l-case": 8,
+    "ADP|l-cc": 9,
+    "ADP|l-mark": 10,
+    "ADP|l-nsubj": 11,
+    "ADP|l-obl": 12,
+    "ADP|r-case": 13,
+    "ADP|r-conj": 14,
+    "ADP|r-fixed": 15,
+    "ADP|r-mark": 16,
+    "ADP|r-obj": 17,
+    "ADP|root": 18,
+    "ADV": 19,
+    "ADV.": 20,
+    "ADV|AdvType=Cau|_": 21,
+    "ADV|AdvType=Cau|l-advmod": 22,
+    "ADV|AdvType=Cau|l-amod": 23,
+    "ADV|AdvType=Cau|l-nsubj": 24,
+    "ADV|AdvType=Cau|l-obj": 25,
+    "ADV|AdvType=Deg|Degree=Cmp|_": 26,
+    "ADV|AdvType=Deg|Degree=Cmp|l-advmod": 27,
+    "ADV|AdvType=Deg|Degree=Cmp|l-amod": 28,
+    "ADV|AdvType=Deg|Degree=Cmp|r-conj": 29,
+    "ADV|AdvType=Deg|Degree=Cmp|r-obj": 30,
+    "ADV|AdvType=Deg|Degree=Pos|_": 31,
+    "ADV|AdvType=Deg|Degree=Pos|l-advmod": 32,
+    "ADV|AdvType=Deg|Degree=Pos|l-amod": 33,
+    "ADV|AdvType=Deg|Degree=Pos|r-ccomp": 34,
+    "ADV|AdvType=Deg|Degree=Pos|r-conj": 35,
+    "ADV|AdvType=Deg|Degree=Pos|r-flat:vv": 36,
+    "ADV|AdvType=Deg|Degree=Pos|r-parataxis": 37,
+    "ADV|AdvType=Deg|Degree=Pos|root": 38,
+    "ADV|AdvType=Deg|Degree=Sup|_": 39,
+    "ADV|AdvType=Deg|Degree=Sup|l-advmod": 40,
+    "ADV|AdvType=Deg|Degree=Sup|l-amod": 41,
+    "ADV|AdvType=Deg|Degree=Sup|l-nsubj": 42,
+    "ADV|AdvType=Deg|Degree=Sup|r-conj": 43,
+    "ADV|AdvType=Deg|Degree=Sup|r-parataxis": 44,
+    "ADV|AdvType=Deg|Degree=Sup|root": 45,
+    "ADV|AdvType=Tim|Aspect=Perf|_": 46,
+    "ADV|AdvType=Tim|Aspect=Perf|l-advmod": 47,
+    "ADV|AdvType=Tim|Aspect=Perf|l-amod": 48,
+    "ADV|AdvType=Tim|Aspect=Perf|l-obl:lmod": 49,
+    "ADV|AdvType=Tim|Aspect=Perf|r-parataxis": 50,
+    "ADV|AdvType=Tim|Aspect=Perf|root": 51,
+    "ADV|AdvType=Tim|Tense=Fut|_": 52,
+    "ADV|AdvType=Tim|Tense=Fut|l-advmod": 53,
+    "ADV|AdvType=Tim|Tense=Fut|l-amod": 54,
+    "ADV|AdvType=Tim|Tense=Fut|l-nsubj": 55,
+    "ADV|AdvType=Tim|Tense=Fut|l-nsubj:outer": 56,
+    "ADV|AdvType=Tim|Tense=Fut|root": 57,
+    "ADV|AdvType=Tim|Tense=Past|_": 58,
+    "ADV|AdvType=Tim|Tense=Past|l-advmod": 59,
+    "ADV|AdvType=Tim|Tense=Past|l-amod": 60,
+    "ADV|AdvType=Tim|Tense=Pres|_": 61,
+    "ADV|AdvType=Tim|Tense=Pres|l-advmod": 62,
+    "ADV|AdvType=Tim|Tense=Pres|l-amod": 63,
+    "ADV|AdvType=Tim|Tense=Pres|root": 64,
+    "ADV|AdvType=Tim|_": 65,
+    "ADV|AdvType=Tim|l-advcl": 66,
+    "ADV|AdvType=Tim|l-advmod": 67,
+    "ADV|AdvType=Tim|l-amod": 68,
+    "ADV|AdvType=Tim|l-nsubj": 69,
+    "ADV|AdvType=Tim|r-advmod": 70,
+    "ADV|AdvType=Tim|r-ccomp": 71,
+    "ADV|AdvType=Tim|r-compound:redup": 72,
+    "ADV|AdvType=Tim|r-conj": 73,
+    "ADV|AdvType=Tim|r-flat:vv": 74,
+    "ADV|AdvType=Tim|r-parataxis": 75,
+    "ADV|AdvType=Tim|root": 76,
+    "ADV|Degree=Equ|VerbForm=Conv|_": 77,
+    "ADV|Degree=Equ|VerbForm=Conv|l-advmod": 78,
+    "ADV|Degree=Pos|VerbForm=Conv|_": 79,
+    "ADV|Degree=Pos|VerbForm=Conv|l-advmod": 80,
+    "ADV|Degree=Pos|VerbForm=Conv|r-advmod": 81,
+    "ADV|Polarity=Neg|VerbForm=Conv|_": 82,
+    "ADV|Polarity=Neg|VerbForm=Conv|l-advmod": 83,
+    "ADV|Polarity=Neg|_": 84,
+    "ADV|Polarity=Neg|l-advmod": 85,
+    "ADV|Polarity=Neg|l-amod": 86,
+    "ADV|Polarity=Neg|l-nsubj": 87,
+    "ADV|Polarity=Neg|l-parataxis": 88,
+    "ADV|Polarity=Neg|r-advmod": 89,
+    "ADV|Polarity=Neg|r-conj": 90,
+    "ADV|Polarity=Neg|r-obj": 91,
+    "ADV|Polarity=Neg|r-parataxis": 92,
+    "ADV|Polarity=Neg|root": 93,
+    "ADV|VerbForm=Conv|_": 94,
+    "ADV|VerbForm=Conv|l-advmod": 95,
+    "ADV|VerbForm=Conv|r-advmod": 96,
+    "ADV|_": 97,
+    "ADV|l-acl": 98,
+    "ADV|l-advcl": 99,
+    "ADV|l-advmod": 100,
+    "ADV|l-amod": 101,
+    "ADV|l-cc": 102,
+    "ADV|l-nsubj": 103,
+    "ADV|r-advmod": 104,
+    "ADV|r-ccomp": 105,
+    "ADV|r-conj": 106,
+    "ADV|r-flat:vv": 107,
+    "ADV|r-obj": 108,
+    "ADV|root": 109,
+    "AUX": 110,
+    "AUX.": 111,
+    "AUX|Mood=Des|_": 112,
+    "AUX|Mood=Des|l-aux": 113,
+    "AUX|Mood=Des|l-csubj": 114,
+    "AUX|Mood=Des|l-parataxis": 115,
+    "AUX|Mood=Des|r-ccomp": 116,
+    "AUX|Mood=Des|r-conj": 117,
+    "AUX|Mood=Des|r-flat:vv": 118,
+    "AUX|Mood=Des|root": 119,
+    "AUX|Mood=Nec|_": 120,
+    "AUX|Mood=Nec|l-acl": 121,
+    "AUX|Mood=Nec|l-amod": 122,
+    "AUX|Mood=Nec|l-aux": 123,
+    "AUX|Mood=Nec|r-aux": 124,
+    "AUX|Mood=Nec|root": 125,
+    "AUX|Mood=Pot|_": 126,
+    "AUX|Mood=Pot|l-acl": 127,
+    "AUX|Mood=Pot|l-advcl": 128,
+    "AUX|Mood=Pot|l-amod": 129,
+    "AUX|Mood=Pot|l-aux": 130,
+    "AUX|Mood=Pot|l-csubj": 131,
+    "AUX|Mood=Pot|l-nsubj": 132,
+    "AUX|Mood=Pot|r-ccomp": 133,
+    "AUX|Mood=Pot|r-conj": 134,
+    "AUX|Mood=Pot|r-obj": 135,
+    "AUX|Mood=Pot|r-parataxis": 136,
+    "AUX|Mood=Pot|r-xcomp": 137,
+    "AUX|Mood=Pot|root": 138,
+    "AUX|VerbType=Cop|_": 139,
+    "AUX|VerbType=Cop|l-cop": 140,
+    "AUX|Voice=Pass|_": 141,
+    "AUX|Voice=Pass|l-aux": 142,
+    "AUX|Voice=Pass|r-conj": 143,
+    "AUX|Voice=Pass|root": 144,
+    "B-ADP": 145,
+    "B-ADP.": 146,
+    "B-ADV": 147,
+    "B-ADV.": 148,
+    "B-AUX": 149,
+    "B-AUX.": 150,
+    "B-CCONJ": 151,
+    "B-CCONJ.": 152,
+    "B-INTJ": 153,
+    "B-INTJ.": 154,
+    "B-NOUN": 155,
+    "B-NOUN.": 156,
+    "B-NUM": 157,
+    "B-NUM.": 158,
+    "B-PART": 159,
+    "B-PART.": 160,
+    "B-PRON": 161,
+    "B-PRON.": 162,
+    "B-PROPN": 163,
+    "B-PROPN.": 164,
+    "B-PUNCT": 165,
+    "B-PUNCT.": 166,
+    "B-SCONJ": 167,
+    "B-SCONJ.": 168,
+    "B-SYM": 169,
+    "B-SYM.": 170,
+    "B-VERB": 171,
+    "B-VERB.": 172,
+    "CCONJ": 173,
+    "CCONJ.": 174,
+    "CCONJ|_": 175,
+    "CCONJ|l-advmod": 176,
+    "CCONJ|l-amod": 177,
+    "CCONJ|l-cc": 178,
+    "CCONJ|l-obj": 179,
+    "CCONJ|r-fixed": 180,
+    "CCONJ|r-orphan": 181,
+    "I-ADP": 182,
+    "I-ADP.": 183,
+    "I-ADV": 184,
+    "I-ADV.": 185,
+    "I-AUX": 186,
+    "I-AUX.": 187,
+    "I-CCONJ": 188,
+    "I-CCONJ.": 189,
+    "I-INTJ": 190,
+    "I-INTJ.": 191,
+    "I-NOUN": 192,
+    "I-NOUN.": 193,
+    "I-NUM": 194,
+    "I-NUM.": 195,
+    "I-PART": 196,
+    "I-PART.": 197,
+    "I-PRON": 198,
+    "I-PRON.": 199,
+    "I-PROPN": 200,
+    "I-PROPN.": 201,
+    "I-PUNCT": 202,
+    "I-PUNCT.": 203,
+    "I-SCONJ": 204,
+    "I-SCONJ.": 205,
+    "I-SYM": 206,
+    "I-SYM.": 207,
+    "I-VERB": 208,
+    "I-VERB.": 209,
+    "INTJ": 210,
+    "INTJ.": 211,
+    "INTJ|_": 212,
+    "INTJ|l-advcl": 213,
+    "INTJ|l-csubj": 214,
+    "INTJ|l-discourse": 215,
+    "INTJ|l-discourse:sp": 216,
+    "INTJ|l-dislocated": 217,
+    "INTJ|l-nsubj": 218,
+    "INTJ|l-vocative": 219,
+    "INTJ|r-compound:redup": 220,
+    "INTJ|r-conj": 221,
+    "INTJ|r-discourse:sp": 222,
+    "INTJ|r-dislocated": 223,
+    "INTJ|r-fixed": 224,
+    "INTJ|r-obj": 225,
+    "INTJ|r-parataxis": 226,
+    "INTJ|root": 227,
+    "NOUN": 228,
+    "NOUN.": 229,
+    "NOUN|Case=Loc|_": 230,
+    "NOUN|Case=Loc|l-acl": 231,
+    "NOUN|Case=Loc|l-advcl": 232,
+    "NOUN|Case=Loc|l-amod": 233,
+    "NOUN|Case=Loc|l-clf": 234,
+    "NOUN|Case=Loc|l-compound": 235,
+    "NOUN|Case=Loc|l-csubj": 236,
+    "NOUN|Case=Loc|l-dislocated": 237,
+    "NOUN|Case=Loc|l-nmod": 238,
+    "NOUN|Case=Loc|l-nsubj": 239,
+    "NOUN|Case=Loc|l-nsubj:outer": 240,
+    "NOUN|Case=Loc|l-obj": 241,
+    "NOUN|Case=Loc|l-obl": 242,
+    "NOUN|Case=Loc|l-obl:lmod": 243,
+    "NOUN|Case=Loc|l-obl:tmod": 244,
+    "NOUN|Case=Loc|l-parataxis": 245,
+    "NOUN|Case=Loc|r-ccomp": 246,
+    "NOUN|Case=Loc|r-clf": 247,
+    "NOUN|Case=Loc|r-compound:redup": 248,
+    "NOUN|Case=Loc|r-conj": 249,
+    "NOUN|Case=Loc|r-dislocated": 250,
+    "NOUN|Case=Loc|r-flat": 251,
+    "NOUN|Case=Loc|r-iobj": 252,
+    "NOUN|Case=Loc|r-list": 253,
+    "NOUN|Case=Loc|r-nmod": 254,
+    "NOUN|Case=Loc|r-nsubj": 255,
+    "NOUN|Case=Loc|r-obj": 256,
+    "NOUN|Case=Loc|r-obl": 257,
+    "NOUN|Case=Loc|r-obl:lmod": 258,
+    "NOUN|Case=Loc|r-parataxis": 259,
+    "NOUN|Case=Loc|r-xcomp": 260,
+    "NOUN|Case=Loc|root": 261,
+    "NOUN|Case=Tem|_": 262,
+    "NOUN|Case=Tem|l-acl": 263,
+    "NOUN|Case=Tem|l-advcl": 264,
+    "NOUN|Case=Tem|l-amod": 265,
+    "NOUN|Case=Tem|l-compound": 266,
+    "NOUN|Case=Tem|l-csubj": 267,
+    "NOUN|Case=Tem|l-nmod": 268,
+    "NOUN|Case=Tem|l-nsubj": 269,
+    "NOUN|Case=Tem|l-nsubj:outer": 270,
+    "NOUN|Case=Tem|l-obj": 271,
+    "NOUN|Case=Tem|l-obl:tmod": 272,
+    "NOUN|Case=Tem|r-amod": 273,
+    "NOUN|Case=Tem|r-ccomp": 274,
+    "NOUN|Case=Tem|r-clf": 275,
+    "NOUN|Case=Tem|r-compound:redup": 276,
+    "NOUN|Case=Tem|r-conj": 277,
+    "NOUN|Case=Tem|r-flat": 278,
+    "NOUN|Case=Tem|r-iobj": 279,
+    "NOUN|Case=Tem|r-list": 280,
+    "NOUN|Case=Tem|r-nsubj": 281,
+    "NOUN|Case=Tem|r-obj": 282,
+    "NOUN|Case=Tem|r-obl:tmod": 283,
+    "NOUN|Case=Tem|r-parataxis": 284,
+    "NOUN|Case=Tem|r-xcomp": 285,
+    "NOUN|Case=Tem|root": 286,
+    "NOUN|Degree=Pos|_": 287,
+    "NOUN|Degree=Pos|root": 288,
+    "NOUN|NounType=Clf|_": 289,
+    "NOUN|NounType=Clf|l-clf": 290,
+    "NOUN|NounType=Clf|l-nmod": 291,
+    "NOUN|NounType=Clf|l-nsubj": 292,
+    "NOUN|NounType=Clf|l-obl": 293,
+    "NOUN|NounType=Clf|r-ccomp": 294,
+    "NOUN|NounType=Clf|r-clf": 295,
+    "NOUN|NounType=Clf|r-compound:redup": 296,
+    "NOUN|NounType=Clf|r-conj": 297,
+    "NOUN|NounType=Clf|r-flat": 298,
+    "NOUN|NounType=Clf|r-obj": 299,
+    "NOUN|NounType=Clf|r-parataxis": 300,
+    "NOUN|NounType=Clf|root": 301,
+    "NOUN|_": 302,
+    "NOUN|l-acl": 303,
+    "NOUN|l-advcl": 304,
+    "NOUN|l-amod": 305,
+    "NOUN|l-ccomp": 306,
+    "NOUN|l-clf": 307,
+    "NOUN|l-compound": 308,
+    "NOUN|l-csubj": 309,
+    "NOUN|l-csubj:outer": 310,
+    "NOUN|l-dislocated": 311,
+    "NOUN|l-iobj": 312,
+    "NOUN|l-list": 313,
+    "NOUN|l-nmod": 314,
+    "NOUN|l-nsubj": 315,
+    "NOUN|l-nsubj:outer": 316,
+    "NOUN|l-nsubj:pass": 317,
+    "NOUN|l-obj": 318,
+    "NOUN|l-obl": 319,
+    "NOUN|l-obl:lmod": 320,
+    "NOUN|l-obl:tmod": 321,
+    "NOUN|l-vocative": 322,
+    "NOUN|r-acl": 323,
+    "NOUN|r-advcl": 324,
+    "NOUN|r-amod": 325,
+    "NOUN|r-ccomp": 326,
+    "NOUN|r-clf": 327,
+    "NOUN|r-compound:redup": 328,
+    "NOUN|r-conj": 329,
+    "NOUN|r-csubj": 330,
+    "NOUN|r-dislocated": 331,
+    "NOUN|r-flat": 332,
+    "NOUN|r-flat:foreign": 333,
+    "NOUN|r-iobj": 334,
+    "NOUN|r-list": 335,
+    "NOUN|r-nmod": 336,
+    "NOUN|r-nsubj": 337,
+    "NOUN|r-obj": 338,
+    "NOUN|r-obl": 339,
+    "NOUN|r-obl:lmod": 340,
+    "NOUN|r-parataxis": 341,
+    "NOUN|r-vocative": 342,
+    "NOUN|r-xcomp": 343,
+    "NOUN|root": 344,
+    "NUM": 345,
+    "NUM.": 346,
+    "NUM|NumType=Ord|_": 347,
+    "NUM|NumType=Ord|l-nsubj": 348,
+    "NUM|NumType=Ord|l-nummod": 349,
+    "NUM|NumType=Ord|l-obl": 350,
+    "NUM|NumType=Ord|l-obl:lmod": 351,
+    "NUM|NumType=Ord|l-obl:tmod": 352,
+    "NUM|NumType=Ord|r-conj": 353,
+    "NUM|NumType=Ord|r-flat": 354,
+    "NUM|NumType=Ord|r-obj": 355,
+    "NUM|NumType=Ord|root": 356,
+    "NUM|_": 357,
+    "NUM|l-acl": 358,
+    "NUM|l-advcl": 359,
+    "NUM|l-compound": 360,
+    "NUM|l-csubj": 361,
+    "NUM|l-dislocated": 362,
+    "NUM|l-nsubj": 363,
+    "NUM|l-nsubj:outer": 364,
+    "NUM|l-nummod": 365,
+    "NUM|l-obj": 366,
+    "NUM|l-obl": 367,
+    "NUM|l-obl:lmod": 368,
+    "NUM|l-obl:tmod": 369,
+    "NUM|r-ccomp": 370,
+    "NUM|r-clf": 371,
+    "NUM|r-compound": 372,
+    "NUM|r-compound:redup": 373,
+    "NUM|r-conj": 374,
+    "NUM|r-flat": 375,
+    "NUM|r-iobj": 376,
+    "NUM|r-list": 377,
+    "NUM|r-nummod": 378,
+    "NUM|r-obj": 379,
+    "NUM|r-obl": 380,
+    "NUM|r-obl:tmod": 381,
+    "NUM|r-parataxis": 382,
+    "NUM|r-xcomp": 383,
+    "NUM|root": 384,
+    "PART": 385,
+    "PART.": 386,
+    "PART|_": 387,
+    "PART|l-acl": 388,
+    "PART|l-advcl": 389,
+    "PART|l-advmod": 390,
+    "PART|l-amod": 391,
+    "PART|l-case": 392,
+    "PART|l-cc": 393,
+    "PART|l-csubj": 394,
+    "PART|l-csubj:outer": 395,
+    "PART|l-discourse": 396,
+    "PART|l-discourse:sp": 397,
+    "PART|l-dislocated": 398,
+    "PART|l-mark": 399,
+    "PART|l-nmod": 400,
+    "PART|l-nsubj": 401,
+    "PART|l-nsubj:outer": 402,
+    "PART|l-nsubj:pass": 403,
+    "PART|l-obj": 404,
+    "PART|l-obl": 405,
+    "PART|l-obl:lmod": 406,
+    "PART|r-advmod": 407,
+    "PART|r-case": 408,
+    "PART|r-ccomp": 409,
+    "PART|r-clf": 410,
+    "PART|r-conj": 411,
+    "PART|r-discourse": 412,
+    "PART|r-discourse:sp": 413,
+    "PART|r-dislocated": 414,
+    "PART|r-fixed": 415,
+    "PART|r-flat": 416,
+    "PART|r-iobj": 417,
+    "PART|r-list": 418,
+    "PART|r-mark": 419,
+    "PART|r-nsubj": 420,
+    "PART|r-obj": 421,
+    "PART|r-obl": 422,
+    "PART|r-parataxis": 423,
+    "PART|r-xcomp": 424,
+    "PART|root": 425,
+    "PRON": 426,
+    "PRON.": 427,
+    "PRON|Person=1|PronType=Prs|_": 428,
+    "PRON|Person=1|PronType=Prs|l-acl": 429,
+    "PRON|Person=1|PronType=Prs|l-advcl": 430,
+    "PRON|Person=1|PronType=Prs|l-det": 431,
+    "PRON|Person=1|PronType=Prs|l-iobj": 432,
+    "PRON|Person=1|PronType=Prs|l-nsubj": 433,
+    "PRON|Person=1|PronType=Prs|l-nsubj:outer": 434,
+    "PRON|Person=1|PronType=Prs|l-obj": 435,
+    "PRON|Person=1|PronType=Prs|l-obl": 436,
+    "PRON|Person=1|PronType=Prs|l-vocative": 437,
+    "PRON|Person=1|PronType=Prs|r-ccomp": 438,
+    "PRON|Person=1|PronType=Prs|r-conj": 439,
+    "PRON|Person=1|PronType=Prs|r-iobj": 440,
+    "PRON|Person=1|PronType=Prs|r-nsubj": 441,
+    "PRON|Person=1|PronType=Prs|r-obj": 442,
+    "PRON|Person=1|PronType=Prs|r-obl": 443,
+    "PRON|Person=1|PronType=Prs|r-obl:lmod": 444,
+    "PRON|Person=1|PronType=Prs|root": 445,
+    "PRON|Person=2|PronType=Prs|_": 446,
+    "PRON|Person=2|PronType=Prs|l-advcl": 447,
+    "PRON|Person=2|PronType=Prs|l-amod": 448,
+    "PRON|Person=2|PronType=Prs|l-det": 449,
+    "PRON|Person=2|PronType=Prs|l-nmod": 450,
+    "PRON|Person=2|PronType=Prs|l-nsubj": 451,
+    "PRON|Person=2|PronType=Prs|l-nsubj:outer": 452,
+    "PRON|Person=2|PronType=Prs|l-obj": 453,
+    "PRON|Person=2|PronType=Prs|l-obl": 454,
+    "PRON|Person=2|PronType=Prs|l-vocative": 455,
+    "PRON|Person=2|PronType=Prs|r-conj": 456,
+    "PRON|Person=2|PronType=Prs|r-flat": 457,
+    "PRON|Person=2|PronType=Prs|r-iobj": 458,
+    "PRON|Person=2|PronType=Prs|r-obj": 459,
+    "PRON|Person=2|PronType=Prs|r-obl": 460,
+    "PRON|Person=2|PronType=Prs|root": 461,
+    "PRON|Person=3|PronType=Prs|_": 462,
+    "PRON|Person=3|PronType=Prs|l-advcl": 463,
+    "PRON|Person=3|PronType=Prs|l-amod": 464,
+    "PRON|Person=3|PronType=Prs|l-det": 465,
+    "PRON|Person=3|PronType=Prs|l-dislocated": 466,
+    "PRON|Person=3|PronType=Prs|l-expl": 467,
+    "PRON|Person=3|PronType=Prs|l-iobj": 468,
+    "PRON|Person=3|PronType=Prs|l-nsubj": 469,
+    "PRON|Person=3|PronType=Prs|l-nsubj:outer": 470,
+    "PRON|Person=3|PronType=Prs|l-nsubj:pass": 471,
+    "PRON|Person=3|PronType=Prs|l-obj": 472,
+    "PRON|Person=3|PronType=Prs|l-obl": 473,
+    "PRON|Person=3|PronType=Prs|r-ccomp": 474,
+    "PRON|Person=3|PronType=Prs|r-conj": 475,
+    "PRON|Person=3|PronType=Prs|r-expl": 476,
+    "PRON|Person=3|PronType=Prs|r-iobj": 477,
+    "PRON|Person=3|PronType=Prs|r-nsubj": 478,
+    "PRON|Person=3|PronType=Prs|r-obj": 479,
+    "PRON|Person=3|PronType=Prs|r-obl": 480,
+    "PRON|Person=3|PronType=Prs|root": 481,
+    "PRON|PronType=Dem|_": 482,
+    "PRON|PronType=Dem|l-acl": 483,
+    "PRON|PronType=Dem|l-advcl": 484,
+    "PRON|PronType=Dem|l-amod": 485,
+    "PRON|PronType=Dem|l-compound": 486,
+    "PRON|PronType=Dem|l-det": 487,
+    "PRON|PronType=Dem|l-dislocated": 488,
+    "PRON|PronType=Dem|l-expl": 489,
+    "PRON|PronType=Dem|l-nsubj": 490,
+    "PRON|PronType=Dem|l-nsubj:outer": 491,
+    "PRON|PronType=Dem|l-obj": 492,
+    "PRON|PronType=Dem|l-obl": 493,
+    "PRON|PronType=Dem|l-obl:lmod": 494,
+    "PRON|PronType=Dem|r-conj": 495,
+    "PRON|PronType=Dem|r-det": 496,
+    "PRON|PronType=Dem|r-expl": 497,
+    "PRON|PronType=Dem|r-flat": 498,
+    "PRON|PronType=Dem|r-iobj": 499,
+    "PRON|PronType=Dem|r-obj": 500,
+    "PRON|PronType=Dem|r-obl": 501,
+    "PRON|PronType=Dem|r-obl:lmod": 502,
+    "PRON|PronType=Dem|root": 503,
+    "PRON|PronType=Int|_": 504,
+    "PRON|PronType=Int|l-advcl": 505,
+    "PRON|PronType=Int|l-amod": 506,
+    "PRON|PronType=Int|l-det": 507,
+    "PRON|PronType=Int|l-dislocated": 508,
+    "PRON|PronType=Int|l-nsubj": 509,
+    "PRON|PronType=Int|l-nsubj:outer": 510,
+    "PRON|PronType=Int|l-obj": 511,
+    "PRON|PronType=Int|l-obl": 512,
+    "PRON|PronType=Int|l-vocative": 513,
+    "PRON|PronType=Int|r-ccomp": 514,
+    "PRON|PronType=Int|r-conj": 515,
+    "PRON|PronType=Int|r-flat": 516,
+    "PRON|PronType=Int|r-obj": 517,
+    "PRON|PronType=Int|r-parataxis": 518,
+    "PRON|PronType=Int|r-xcomp": 519,
+    "PRON|PronType=Int|root": 520,
+    "PRON|PronType=Prs|Reflex=Yes|_": 521,
+    "PRON|PronType=Prs|Reflex=Yes|l-acl": 522,
+    "PRON|PronType=Prs|Reflex=Yes|l-det": 523,
+    "PRON|PronType=Prs|Reflex=Yes|l-nsubj": 524,
+    "PRON|PronType=Prs|Reflex=Yes|l-obj": 525,
+    "PRON|PronType=Prs|Reflex=Yes|l-obl": 526,
+    "PRON|PronType=Prs|Reflex=Yes|r-dislocated": 527,
+    "PRON|PronType=Prs|Reflex=Yes|r-obj": 528,
+    "PRON|PronType=Prs|Reflex=Yes|r-obl": 529,
+    "PRON|PronType=Prs|Reflex=Yes|root": 530,
+    "PRON|PronType=Prs|_": 531,
+    "PRON|PronType=Prs|l-det": 532,
+    "PRON|PronType=Prs|l-nsubj": 533,
+    "PRON|PronType=Prs|l-nsubj:outer": 534,
+    "PRON|PronType=Prs|l-obj": 535,
+    "PRON|PronType=Prs|r-conj": 536,
+    "PRON|PronType=Prs|r-iobj": 537,
+    "PRON|PronType=Prs|r-obj": 538,
+    "PROPN": 539,
+    "PROPN.": 540,
+    "PROPN|Case=Loc|NameType=Geo|_": 541,
+    "PROPN|Case=Loc|NameType=Geo|l-acl": 542,
+    "PROPN|Case=Loc|NameType=Geo|l-advcl": 543,
+    "PROPN|Case=Loc|NameType=Geo|l-amod": 544,
+    "PROPN|Case=Loc|NameType=Geo|l-compound": 545,
+    "PROPN|Case=Loc|NameType=Geo|l-csubj": 546,
+    "PROPN|Case=Loc|NameType=Geo|l-dislocated": 547,
+    "PROPN|Case=Loc|NameType=Geo|l-nmod": 548,
+    "PROPN|Case=Loc|NameType=Geo|l-nsubj": 549,
+    "PROPN|Case=Loc|NameType=Geo|l-nsubj:outer": 550,
+    "PROPN|Case=Loc|NameType=Geo|l-obl": 551,
+    "PROPN|Case=Loc|NameType=Geo|l-obl:lmod": 552,
+    "PROPN|Case=Loc|NameType=Geo|r-conj": 553,
+    "PROPN|Case=Loc|NameType=Geo|r-flat": 554,
+    "PROPN|Case=Loc|NameType=Geo|r-iobj": 555,
+    "PROPN|Case=Loc|NameType=Geo|r-obj": 556,
+    "PROPN|Case=Loc|NameType=Geo|r-obl": 557,
+    "PROPN|Case=Loc|NameType=Geo|r-obl:lmod": 558,
+    "PROPN|Case=Loc|NameType=Geo|r-parataxis": 559,
+    "PROPN|Case=Loc|NameType=Geo|r-xcomp": 560,
+    "PROPN|Case=Loc|NameType=Geo|root": 561,
+    "PROPN|Case=Loc|NameType=Nat|_": 562,
+    "PROPN|Case=Loc|NameType=Nat|l-acl": 563,
+    "PROPN|Case=Loc|NameType=Nat|l-advcl": 564,
+    "PROPN|Case=Loc|NameType=Nat|l-amod": 565,
+    "PROPN|Case=Loc|NameType=Nat|l-clf": 566,
+    "PROPN|Case=Loc|NameType=Nat|l-compound": 567,
+    "PROPN|Case=Loc|NameType=Nat|l-nmod": 568,
+    "PROPN|Case=Loc|NameType=Nat|l-nsubj": 569,
+    "PROPN|Case=Loc|NameType=Nat|l-nsubj:outer": 570,
+    "PROPN|Case=Loc|NameType=Nat|l-nsubj:pass": 571,
+    "PROPN|Case=Loc|NameType=Nat|l-obj": 572,
+    "PROPN|Case=Loc|NameType=Nat|l-obl": 573,
+    "PROPN|Case=Loc|NameType=Nat|l-obl:lmod": 574,
+    "PROPN|Case=Loc|NameType=Nat|r-ccomp": 575,
+    "PROPN|Case=Loc|NameType=Nat|r-conj": 576,
+    "PROPN|Case=Loc|NameType=Nat|r-flat": 577,
+    "PROPN|Case=Loc|NameType=Nat|r-iobj": 578,
+    "PROPN|Case=Loc|NameType=Nat|r-nmod": 579,
+    "PROPN|Case=Loc|NameType=Nat|r-obj": 580,
+    "PROPN|Case=Loc|NameType=Nat|r-obl": 581,
+    "PROPN|Case=Loc|NameType=Nat|r-obl:lmod": 582,
+    "PROPN|Case=Loc|NameType=Nat|r-parataxis": 583,
+    "PROPN|Case=Loc|NameType=Nat|r-xcomp": 584,
+    "PROPN|Case=Loc|NameType=Nat|root": 585,
+    "PROPN|NameType=Giv|_": 586,
+    "PROPN|NameType=Giv|l-acl": 587,
+    "PROPN|NameType=Giv|l-advcl": 588,
+    "PROPN|NameType=Giv|l-amod": 589,
+    "PROPN|NameType=Giv|l-compound": 590,
+    "PROPN|NameType=Giv|l-dislocated": 591,
+    "PROPN|NameType=Giv|l-nmod": 592,
+    "PROPN|NameType=Giv|l-nsubj": 593,
+    "PROPN|NameType=Giv|l-nsubj:outer": 594,
+    "PROPN|NameType=Giv|l-nsubj:pass": 595,
+    "PROPN|NameType=Giv|l-obj": 596,
+    "PROPN|NameType=Giv|l-obl": 597,
+    "PROPN|NameType=Giv|l-obl:lmod": 598,
+    "PROPN|NameType=Giv|l-parataxis": 599,
+    "PROPN|NameType=Giv|l-vocative": 600,
+    "PROPN|NameType=Giv|r-appos": 601,
+    "PROPN|NameType=Giv|r-ccomp": 602,
+    "PROPN|NameType=Giv|r-conj": 603,
+    "PROPN|NameType=Giv|r-dislocated": 604,
+    "PROPN|NameType=Giv|r-flat": 605,
+    "PROPN|NameType=Giv|r-iobj": 606,
+    "PROPN|NameType=Giv|r-list": 607,
+    "PROPN|NameType=Giv|r-nmod": 608,
+    "PROPN|NameType=Giv|r-obj": 609,
+    "PROPN|NameType=Giv|r-obl": 610,
+    "PROPN|NameType=Giv|r-obl:lmod": 611,
+    "PROPN|NameType=Giv|r-parataxis": 612,
+    "PROPN|NameType=Giv|r-xcomp": 613,
+    "PROPN|NameType=Giv|root": 614,
+    "PROPN|NameType=Prs|_": 615,
+    "PROPN|NameType=Prs|l-acl": 616,
+    "PROPN|NameType=Prs|l-advcl": 617,
+    "PROPN|NameType=Prs|l-amod": 618,
+    "PROPN|NameType=Prs|l-compound": 619,
+    "PROPN|NameType=Prs|l-dislocated": 620,
+    "PROPN|NameType=Prs|l-nmod": 621,
+    "PROPN|NameType=Prs|l-nsubj": 622,
+    "PROPN|NameType=Prs|l-nsubj:outer": 623,
+    "PROPN|NameType=Prs|l-obj": 624,
+    "PROPN|NameType=Prs|l-obl": 625,
+    "PROPN|NameType=Prs|r-conj": 626,
+    "PROPN|NameType=Prs|r-dislocated": 627,
+    "PROPN|NameType=Prs|r-flat": 628,
+    "PROPN|NameType=Prs|r-iobj": 629,
+    "PROPN|NameType=Prs|r-obj": 630,
+    "PROPN|NameType=Prs|r-obl": 631,
+    "PROPN|NameType=Prs|r-parataxis": 632,
+    "PROPN|NameType=Prs|root": 633,
+    "PROPN|NameType=Sur|_": 634,
+    "PROPN|NameType=Sur|l-acl": 635,
+    "PROPN|NameType=Sur|l-advcl": 636,
+    "PROPN|NameType=Sur|l-amod": 637,
+    "PROPN|NameType=Sur|l-compound": 638,
+    "PROPN|NameType=Sur|l-csubj": 639,
+    "PROPN|NameType=Sur|l-dislocated": 640,
+    "PROPN|NameType=Sur|l-nmod": 641,
+    "PROPN|NameType=Sur|l-nsubj": 642,
+    "PROPN|NameType=Sur|l-nsubj:outer": 643,
+    "PROPN|NameType=Sur|l-nsubj:pass": 644,
+    "PROPN|NameType=Sur|l-obl": 645,
+    "PROPN|NameType=Sur|l-obl:lmod": 646,
+    "PROPN|NameType=Sur|l-vocative": 647,
+    "PROPN|NameType=Sur|r-ccomp": 648,
+    "PROPN|NameType=Sur|r-conj": 649,
+    "PROPN|NameType=Sur|r-dislocated": 650,
+    "PROPN|NameType=Sur|r-flat": 651,
+    "PROPN|NameType=Sur|r-iobj": 652,
+    "PROPN|NameType=Sur|r-list": 653,
+    "PROPN|NameType=Sur|r-nmod": 654,
+    "PROPN|NameType=Sur|r-nsubj": 655,
+    "PROPN|NameType=Sur|r-obj": 656,
+    "PROPN|NameType=Sur|r-obl": 657,
+    "PROPN|NameType=Sur|r-obl:lmod": 658,
+    "PROPN|NameType=Sur|r-parataxis": 659,
+    "PROPN|NameType=Sur|r-xcomp": 660,
+    "PROPN|NameType=Sur|root": 661,
+    "PROPN|_": 662,
+    "PROPN|l-nmod": 663,
+    "PUNCT": 664,
+    "PUNCT.": 665,
+    "PUNCT|_": 666,
+    "PUNCT|root": 667,
+    "SCONJ": 668,
+    "SCONJ.": 669,
+    "SCONJ|_": 670,
+    "SCONJ|l-case": 671,
+    "SCONJ|l-cc": 672,
+    "SCONJ|l-mark": 673,
+    "SCONJ|l-nsubj": 674,
+    "SCONJ|l-obl": 675,
+    "SCONJ|r-case": 676,
+    "SCONJ|r-iobj": 677,
+    "SCONJ|r-mark": 678,
+    "SCONJ|r-nsubj": 679,
+    "SCONJ|r-nsubj:pass": 680,
+    "SCONJ|r-obj": 681,
+    "SCONJ|root": 682,
+    "SYM": 683,
+    "SYM.": 684,
+    "SYM|_": 685,
+    "SYM|l-nmod": 686,
+    "SYM|l-nsubj": 687,
+    "SYM|r-conj": 688,
+    "SYM|r-nmod": 689,
+    "SYM|r-xcomp": 690,
+    "SYM|root": 691,
+    "VERB": 692,
+    "VERB.": 693,
+    "VERB|Degree=Equ|VerbForm=Part|_": 694,
+    "VERB|Degree=Equ|VerbForm=Part|l-amod": 695,
+    "VERB|Degree=Equ|_": 696,
+    "VERB|Degree=Equ|l-acl": 697,
+    "VERB|Degree=Equ|l-advcl": 698,
+    "VERB|Degree=Equ|l-ccomp": 699,
+    "VERB|Degree=Equ|l-csubj": 700,
+    "VERB|Degree=Equ|l-nsubj": 701,
+    "VERB|Degree=Equ|l-obj": 702,
+    "VERB|Degree=Equ|r-ccomp": 703,
+    "VERB|Degree=Equ|r-compound:redup": 704,
+    "VERB|Degree=Equ|r-conj": 705,
+    "VERB|Degree=Equ|r-obj": 706,
+    "VERB|Degree=Equ|r-parataxis": 707,
+    "VERB|Degree=Equ|r-xcomp": 708,
+    "VERB|Degree=Equ|root": 709,
+    "VERB|Degree=Pos|VerbForm=Part|_": 710,
+    "VERB|Degree=Pos|VerbForm=Part|l-amod": 711,
+    "VERB|Degree=Pos|VerbForm=Part|r-amod": 712,
+    "VERB|Degree=Pos|_": 713,
+    "VERB|Degree=Pos|l-acl": 714,
+    "VERB|Degree=Pos|l-advcl": 715,
+    "VERB|Degree=Pos|l-ccomp": 716,
+    "VERB|Degree=Pos|l-csubj": 717,
+    "VERB|Degree=Pos|l-csubj:outer": 718,
+    "VERB|Degree=Pos|l-dislocated": 719,
+    "VERB|Degree=Pos|l-nsubj": 720,
+    "VERB|Degree=Pos|l-nsubj:outer": 721,
+    "VERB|Degree=Pos|l-obj": 722,
+    "VERB|Degree=Pos|l-obl": 723,
+    "VERB|Degree=Pos|l-vocative": 724,
+    "VERB|Degree=Pos|r-advcl": 725,
+    "VERB|Degree=Pos|r-ccomp": 726,
+    "VERB|Degree=Pos|r-compound:redup": 727,
+    "VERB|Degree=Pos|r-conj": 728,
+    "VERB|Degree=Pos|r-dislocated": 729,
+    "VERB|Degree=Pos|r-fixed": 730,
+    "VERB|Degree=Pos|r-flat:vv": 731,
+    "VERB|Degree=Pos|r-iobj": 732,
+    "VERB|Degree=Pos|r-obj": 733,
+    "VERB|Degree=Pos|r-obl": 734,
+    "VERB|Degree=Pos|r-parataxis": 735,
+    "VERB|Degree=Pos|r-xcomp": 736,
+    "VERB|Degree=Pos|root": 737,
+    "VERB|Polarity=Neg|VerbForm=Part|_": 738,
+    "VERB|Polarity=Neg|VerbForm=Part|l-amod": 739,
+    "VERB|Polarity=Neg|_": 740,
+    "VERB|Polarity=Neg|l-acl": 741,
+    "VERB|Polarity=Neg|l-advcl": 742,
+    "VERB|Polarity=Neg|l-ccomp": 743,
+    "VERB|Polarity=Neg|l-csubj": 744,
+    "VERB|Polarity=Neg|l-csubj:outer": 745,
+    "VERB|Polarity=Neg|l-nsubj": 746,
+    "VERB|Polarity=Neg|l-obl": 747,
+    "VERB|Polarity=Neg|r-advcl": 748,
+    "VERB|Polarity=Neg|r-ccomp": 749,
+    "VERB|Polarity=Neg|r-conj": 750,
+    "VERB|Polarity=Neg|r-flat:vv": 751,
+    "VERB|Polarity=Neg|r-obj": 752,
+    "VERB|Polarity=Neg|r-obl": 753,
+    "VERB|Polarity=Neg|r-parataxis": 754,
+    "VERB|Polarity=Neg|r-xcomp": 755,
+    "VERB|Polarity=Neg|root": 756,
+    "VERB|VerbForm=Part|_": 757,
+    "VERB|VerbForm=Part|l-amod": 758,
+    "VERB|VerbForm=Part|r-amod": 759,
+    "VERB|_": 760,
+    "VERB|l-acl": 761,
+    "VERB|l-advcl": 762,
+    "VERB|l-ccomp": 763,
+    "VERB|l-csubj": 764,
+    "VERB|l-csubj:outer": 765,
+    "VERB|l-csubj:pass": 766,
+    "VERB|l-dislocated": 767,
+    "VERB|l-nsubj": 768,
+    "VERB|l-nsubj:outer": 769,
+    "VERB|l-obj": 770,
+    "VERB|l-obl": 771,
+    "VERB|l-obl:lmod": 772,
+    "VERB|l-parataxis": 773,
+    "VERB|r-acl": 774,
+    "VERB|r-advcl": 775,
+    "VERB|r-ccomp": 776,
+    "VERB|r-compound:redup": 777,
+    "VERB|r-conj": 778,
+    "VERB|r-dislocated": 779,
+    "VERB|r-fixed": 780,
+    "VERB|r-flat:vv": 781,
+    "VERB|r-iobj": 782,
+    "VERB|r-list": 783,
+    "VERB|r-obj": 784,
+    "VERB|r-obl": 785,
+    "VERB|r-obl:lmod": 786,
+    "VERB|r-parataxis": 787,
+    "VERB|r-vocative": 788,
+    "VERB|r-xcomp": 789,
+    "VERB|root": 790
+  },
+  "layer_norm_eps": 1e-05,
+  "local_attention": 128,
+  "local_rope_theta": 10000.0,
+  "max_position_embeddings": 8192,
+  "mlp_bias": false,
+  "mlp_dropout": 0.0,
+  "model_type": "modernbert",
+  "norm_bias": false,
+  "norm_eps": 1e-05,
+  "num_attention_heads": 12,
+  "num_hidden_layers": 22,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "reference_compile": false,
+  "repad_logits_with_grad": false,
+  "sep_token_id": 2,
+  "sparse_pred_ignore_index": -100,
+  "sparse_prediction": false,
+  "tokenizer_class": "BertTokenizerFast",
+  "torch_dtype": "float32",
+  "transformers_version": "4.48.3",
+  "vocab_size": 25078
+}

maker.py

@@ -0,0 +1,118 @@
+#! /usr/bin/python3
+src="KoichiYasuoka/modernbert-base-classical-chinese"
+tgt="KoichiYasuoka/modernbert-base-classical-chinese-ud-embeds"
+url="https://github.com/UniversalDependencies/UD_Classical_Chinese-Kyoto"
+import os
+d=os.path.basename(url)
+os.system("test -d "+d+" || git clone --depth=1 "+url)
+os.system("for F in train dev test ; do cp "+d+"/*-$F.conllu $F.conllu ; done")
+class UDEmbedsDataset(object):
+  def __init__(self,conllu,tokenizer,embeddings=None):
+    self.conllu=open(conllu,"r",encoding="utf-8")
+    self.tokenizer=tokenizer
+    self.embeddings=embeddings
+    self.seeks=[0]
+    label=set(["SYM","SYM.","SYM|_"])
+    dep=set()
+    s=self.conllu.readline()
+    while s!="":
+      if s=="\n":
+        self.seeks.append(self.conllu.tell())
+      else:
+        w=s.split("\t")
+        if len(w)==10:
+          if w[0].isdecimal():
+            p=w[3]
+            q="" if w[5]=="_" else "|"+w[5]
+            d=("|" if w[6]=="0" else "|l-" if int(w[0])<int(w[6]) else "|r-")+w[7]
+            for k in [p,p+".","B-"+p,"B-"+p+".","I-"+p,"I-"+p+".",p+q+"|_",p+q+d]:
+              label.add(k)
+      s=self.conllu.readline()
+    self.label2id={l:i for i,l in enumerate(sorted(label))}
+  def __call__(*args):
+    lid={l:i for i,l in enumerate(sorted(set(sum([list(t.label2id) for t in args],[]))))}
+    for t in args:
+      t.label2id=lid
+    return lid
+  def __del__(self):
+    self.conllu.close()
+  __len__=lambda self:(len(self.seeks)-1)*2
+  def __getitem__(self,i):
+    self.conllu.seek(self.seeks[int(i/2)])
+    z,c,t,s=i%2,[],[""],False
+    while t[0]!="\n":
+      t=self.conllu.readline().split("\t") 
+      if len(t)==10 and t[0].isdecimal():
+        if s:
+          t[1]=" "+t[1]
+        c.append(t)
+        s=t[9].find("SpaceAfter=No")<0
+    x=[True if t[6]=="0" or int(t[6])>j or sum([1 if int(c[i][6])==j+1 else 0 for i in range(j+1,len(c))])>0 else False for j,t in enumerate(c)]
+    v=self.tokenizer([t[1] for t in c],add_special_tokens=False)["input_ids"]
+    if z==0:
+      ids,upos=[self.tokenizer.cls_token_id],["SYM."]
+      for i,(j,k) in enumerate(zip(v,c)):
+        if j==[]:
+          j=[self.tokenizer.unk_token_id]
+        p=k[3] if x[i] else k[3]+"."
+        ids+=j
+        upos+=[p] if len(j)==1 else ["B-"+p]+["I-"+p]*(len(j)-1)
+      ids.append(self.tokenizer.sep_token_id)
+      upos.append("SYM.")
+      emb=self.embeddings
+    else:
+      import torch
+      if len(x)<127:
+        x=[True]*len(x)
+        w=(len(x)+1)*(len(x)+2)/2
+      else:
+        w=sum([len(x)-i+1 if b else 0 for i,b in enumerate(x)])+1
+        for i in range(len(x)):
+          if x[i]==False and w+len(x)-i<8192:
+            x[i]=True
+            w+=len(x)-i+1
+      p=[t[3] if t[5]=="_" else t[3]+"|"+t[5] for i,t in enumerate(c)]
+      d=[t[7] if t[6]=="0" else "l-"+t[7] if int(t[0])<int(t[6]) else "r-"+t[7] for t in c]
+      ids,upos=[-1],["SYM|_"]
+      for i in range(len(x)):
+        if x[i]:
+          ids.append(i)
+          upos.append(p[i]+"|"+d[i] if c[i][6]=="0" else p[i]+"|_")
+          for j in range(i+1,len(x)):
+            ids.append(j)
+            upos.append(p[j]+"|"+d[j] if int(c[j][6])==i+1 else p[i]+"|"+d[i] if int(c[i][6])==j+1 else p[j]+"|_")
+          if i>0 and w>8192:
+            while w>8192:
+              if upos[-1].endswith("|_"):
+                upos.pop(-1)
+                ids.pop(-1)
+                w-=1
+              else:
+                break
+          ids.append(-1)
+          upos.append("SYM|_")
+      with torch.no_grad():
+        m=[]
+        for j in v:
+          if j==[]:
+            j=[self.tokenizer.unk_token_id]
+          m.append(self.embeddings[j,:].sum(axis=0))
+        m.append(self.embeddings[self.tokenizer.sep_token_id,:])
+        emb=torch.stack(m)
+    return{"inputs_embeds":emb[ids,:],"labels":[self.label2id[p] for p in upos]}
+from transformers import AutoTokenizer,AutoConfig,AutoModelForTokenClassification,DefaultDataCollator,TrainingArguments,Trainer
+from tokenizers.pre_tokenizers import Sequence,Split
+from tokenizers import Regex
+tkz=AutoTokenizer.from_pretrained(src)
+trainDS=UDEmbedsDataset("train.conllu",tkz)
+devDS=UDEmbedsDataset("dev.conllu",tkz)
+testDS=UDEmbedsDataset("test.conllu",tkz)
+lid=trainDS(devDS,testDS)
+cfg=AutoConfig.from_pretrained(src,num_labels=len(lid),label2id=lid,id2label={i:l for l,i in lid.items()},ignore_mismatched_sizes=True,trust_remote_code=True)
+mdl=AutoModelForTokenClassification.from_pretrained(src,config=cfg,ignore_mismatched_sizes=True,trust_remote_code=True)
+trainDS.embeddings=mdl.get_input_embeddings().weight
+arg=TrainingArguments(num_train_epochs=10,per_device_train_batch_size=1,dataloader_pin_memory=False,output_dir=tgt,overwrite_output_dir=True,save_total_limit=2,learning_rate=5e-05,warmup_ratio=0.1,save_safetensors=False)
+trn=Trainer(args=arg,data_collator=DefaultDataCollator(),model=mdl,train_dataset=trainDS)
+trn.train()
+trn.save_model(tgt)
+tkz.save_pretrained(tgt)

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d7731f1b2b387efddefea41272e142956c59e06444b8529c88076ab81e77b935
+size 523207026

special_tokens_map.json

@@ -0,0 +1,37 @@
+{
+  "cls_token": {
+    "content": "[CLS]",
+    "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,
+    "normalized": false,
+    "rstrip": false,
+    "single_word": false
+  }
+}

tokenizer_config.json

@@ -0,0 +1,71 @@
+{
+  "added_tokens_decoder": {
+    "0": {
+      "content": "[CLS]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "1": {
+      "content": "[PAD]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "2": {
+      "content": "[SEP]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "3": {
+      "content": "[UNK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    },
+    "4": {
+      "content": "[MASK]",
+      "lstrip": false,
+      "normalized": false,
+      "rstrip": false,
+      "single_word": false,
+      "special": true
+    }
+  },
+  "clean_up_tokenization_spaces": true,
+  "cls_token": "[CLS]",
+  "do_basic_tokenize": true,
+  "do_lower_case": false,
+  "extra_special_tokens": {},
+  "mask_token": "[MASK]",
+  "model_input_names": [
+    "input_ids",
+    "attention_mask"
+  ],
+  "model_max_length": 1000000000000000019884624838656,
+  "never_split": [
+    "[CLS]",
+    "[PAD]",
+    "[SEP]",
+    "[UNK]",
+    "[MASK]"
+  ],
+  "pad_token": "[PAD]",
+  "sep_token": "[SEP]",
+  "stride": 0,
+  "strip_accents": false,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizerFast",
+  "truncation_side": "right",
+  "truncation_strategy": "longest_first",
+  "unk_token": "[UNK]"
+}

ud.py

@@ -0,0 +1,150 @@
+import numpy
+from transformers import TokenClassificationPipeline
+
+class BellmanFordTokenClassificationPipeline(TokenClassificationPipeline):
+  def __init__(self,**kwargs):
+    super().__init__(**kwargs)
+    x=self.model.config.label2id
+    y=[k for k in x if k.find("|")<0 and not k.startswith("I-")]
+    self.transition=numpy.full((len(x),len(x)),-numpy.inf)
+    for k,v in x.items():
+      if k.find("|")<0:
+        for j in ["I-"+k[2:]] if k.startswith("B-") else [k]+y if k.startswith("I-") else y:
+          self.transition[v,x[j]]=0
+  def check_model_type(self,supported_models):
+    pass
+  def postprocess(self,model_outputs,**kwargs):
+    if "logits" not in model_outputs:
+      return self.postprocess(model_outputs[0],**kwargs)
+    return self.bellman_ford_token_classification(model_outputs,**kwargs)
+  def bellman_ford_token_classification(self,model_outputs,**kwargs):
+    m=model_outputs["logits"][0].numpy()
+    e=numpy.exp(m-numpy.max(m,axis=-1,keepdims=True))
+    z=e/e.sum(axis=-1,keepdims=True)
+    for i in range(m.shape[0]-1,0,-1):
+      m[i-1]+=numpy.max(m[i]+self.transition,axis=1)
+    k=[numpy.argmax(m[0]+self.transition[0])]
+    for i in range(1,m.shape[0]):
+      k.append(numpy.argmax(m[i]+self.transition[k[-1]]))
+    w=[{"entity":self.model.config.id2label[j],"start":s,"end":e,"score":z[i,j]} for i,((s,e),j) in enumerate(zip(model_outputs["offset_mapping"][0].tolist(),k)) if s<e]
+    if "aggregation_strategy" in kwargs and kwargs["aggregation_strategy"]!="none":
+      for i,t in reversed(list(enumerate(w))):
+        p=t.pop("entity")
+        if p.startswith("I-"):
+          w[i-1]["score"]=min(w[i-1]["score"],t["score"])
+          w[i-1]["end"]=w.pop(i)["end"]
+        elif p.startswith("B-"):
+          t["entity_group"]=p[2:]
+        else:
+          t["entity_group"]=p
+    for t in w:
+      t["text"]=model_outputs["sentence"][t["start"]:t["end"]]
+    return w
+
+class UniversalDependenciesPipeline(BellmanFordTokenClassificationPipeline):
+  def __init__(self,**kwargs):
+    kwargs["aggregation_strategy"]="simple"
+    super().__init__(**kwargs)
+    x=self.model.config.label2id
+    self.root=numpy.full((len(x)),-numpy.inf)
+    self.left_arc=numpy.full((len(x)),-numpy.inf)
+    self.right_arc=numpy.full((len(x)),-numpy.inf)
+    for k,v in x.items():
+      if k.endswith("|root"):
+        self.root[v]=0
+      elif k.find("|l-")>0:
+        self.left_arc[v]=0
+      elif k.find("|r-")>0:
+        self.right_arc[v]=0
+  def postprocess(self,model_outputs,**kwargs):
+    import torch
+    kwargs["aggregation_strategy"]="simple"
+    if "logits" not in model_outputs:
+      return self.postprocess(model_outputs[0],**kwargs)
+    w=self.bellman_ford_token_classification(model_outputs,**kwargs)
+    off=[(t["start"],t["end"]) for t in w]
+    for i,(s,e) in reversed(list(enumerate(off))):
+      if s<e:
+        d=w[i]["text"]
+        j=len(d)-len(d.lstrip())
+        if j>0:
+          d=d.lstrip()
+          off[i]=(off[i][0]+j,off[i][1])
+        j=len(d)-len(d.rstrip())
+        if j>0:
+          d=d.rstrip()
+          off[i]=(off[i][0],off[i][1]-j)
+        if d.strip()=="":
+          off.pop(i)
+          w.pop(i)
+    v=self.tokenizer([t["text"] for t in w],add_special_tokens=False)
+    x=[not t["entity_group"].endswith(".") for t in w]
+    if len(x)<127:
+      x=[True]*len(x)
+    else:
+      k=sum([len(x)-i+1 if b else 0 for i,b in enumerate(x)])+1
+      for i in numpy.argsort(numpy.array([t["score"] for t in w])):
+        if x[i]==False and k+len(x)-i<8192:
+          x[i]=True
+          k+=len(x)-i+1
+    ids=[-1]
+    for i in range(len(x)):
+      if x[i]:
+        ids.append(i)
+        for j in range(i+1,len(x)):
+          ids.append(j)
+        ids.append(-1)
+    with torch.no_grad():
+      e=self.model.get_input_embeddings().weight
+      m=[]
+      for j in v["input_ids"]:
+        if j==[]:
+          j=[self.tokenizer.unk_token_id]
+        m.append(e[j,:].sum(axis=0))
+      m.append(e[self.tokenizer.sep_token_id,:])
+      m=torch.stack(m).to(self.device)
+      e=self.model(inputs_embeds=torch.unsqueeze(m[ids,:],0))
+    m=e.logits[0].cpu().numpy()
+    e=numpy.full((len(x),len(x),m.shape[-1]),m.min())
+    k=1
+    for i in range(len(x)):
+      if x[i]:
+        e[i,i]=m[k]+self.root
+        k+=1
+        for j in range(1,len(x)-i):
+          e[i+j,i]=m[k]+self.left_arc
+          e[i,i+j]=m[k]+self.right_arc
+          k+=1
+        k+=1
+    m,p=numpy.max(e,axis=2),numpy.argmax(e,axis=2)
+    h=self.chu_liu_edmonds(m)
+    z=[i for i,j in enumerate(h) if i==j]
+    if len(z)>1:
+      k,h=z[numpy.argmax(m[z,z])],numpy.min(m)-numpy.max(m)
+      m[:,z]+=[[0 if j in z and (i!=j or i==k) else h for i in z] for j in range(m.shape[0])]
+      h=self.chu_liu_edmonds(m)
+    q=[self.model.config.id2label[p[j,i]].split("|") for i,j in enumerate(h)]
+    t=model_outputs["sentence"].replace("\n"," ")
+    u="# text = "+t+"\n"
+    for i,(s,e) in enumerate(off):
+      u+="\t".join([str(i+1),t[s:e],"_",q[i][0],"_","_" if len(q[i])<3 else "|".join(q[i][1:-1]),str(0 if h[i]==i else h[i]+1),"root" if q[i][-1]=="root" else q[i][-1][2:],"_","_" if i+1<len(off) and e<off[i+1][0] else "SpaceAfter=No"])+"\n"
+    return u+"\n"
+  def chu_liu_edmonds(self,matrix):
+    h=numpy.argmax(matrix,axis=0)
+    x=[-1 if i==j else j for i,j in enumerate(h)]
+    for b in [lambda x,i,j:-1 if i not in x else x[i],lambda x,i,j:-1 if j<0 else x[j]]:
+      y=[]
+      while x!=y:
+        y=list(x)
+        for i,j in enumerate(x):
+          x[i]=b(x,i,j)
+      if max(x)<0:
+        return h
+    y,x=[i for i,j in enumerate(x) if j==max(x)],[i for i,j in enumerate(x) if j<max(x)]
+    z=matrix-numpy.max(matrix,axis=0)
+    m=numpy.block([[z[x,:][:,x],numpy.max(z[x,:][:,y],axis=1).reshape(len(x),1)],[numpy.max(z[y,:][:,x],axis=0),numpy.max(z[y,y])]])
+    k=[j if i==len(x) else x[j] if j<len(x) else y[numpy.argmax(z[y,x[i]])] for i,j in enumerate(self.chu_liu_edmonds(m))]
+    h=[j if i in y else k[x.index(i)] for i,j in enumerate(h)]
+    i=y[numpy.argmax(z[x[k[-1]],y] if k[-1]<len(x) else z[y,y])]
+    h[i]=x[k[-1]] if k[-1]<len(x) else i
+    return h