initial release

README.md

@@ -0,0 +1,76 @@
+---
+language:
+- "th"
+tags:
+- "thai"
+- "token-classification"
+- "pos"
+- "dependency-parsing"
+datasets:
+- "universal_dependencies"
+license: "apache-2.0"
+pipeline_tag: "token-classification"
+widget:
+- text: "หลายหัวดีกว่าหัวเดียว"
+---
+
+# roberta-base-thai-syllable-ud-goeswith
+
+## Model Description
+
+This is a RoBERTa model pre-trained on Thai Wikipedia texts for POS-tagging and dependency-parsing (using `goeswith` for subwords), derived from [roberta-base-thai-syllable](https://huggingface.co/KoichiYasuoka/roberta-base-thai-syllable).
+
+## How to Use
+
+```py
+class UDgoeswith(object):
+  def __init__(self,bert):
+    from transformers import AutoTokenizer,AutoModelForTokenClassification
+    self.tokenizer=AutoTokenizer.from_pretrained(bert)
+    self.model=AutoModelForTokenClassification.from_pretrained(bert)
+  def __call__(self,text):
+    import numpy,torch,ufal.chu_liu_edmonds
+    w=self.tokenizer(text,return_offsets_mapping=True)
+    v=w["input_ids"]
+    x=[v[0:i]+[self.tokenizer.mask_token_id]+v[i+1:]+[j] for i,j in enumerate(v[1:-1],1)]
+    with torch.no_grad():
+      e=self.model(input_ids=torch.tensor(x)).logits.numpy()[:,1:-2,:]
+    r=[1 if i==0 else -1 if j.endswith("|root") else 0 for i,j in sorted(self.model.config.id2label.items())]
+    e+=numpy.where(numpy.add.outer(numpy.identity(e.shape[0]),r)==0,0,numpy.nan)
+    g=self.model.config.label2id["X|_|goeswith"]
+    r=numpy.tri(e.shape[0])
+    for i in range(e.shape[0]):
+      for j in range(i+2,e.shape[1]):
+        r[i,j]=r[i,j-1] if numpy.nanargmax(e[i,j-1])==g else 1
+    e[:,:,g]+=numpy.where(r==0,0,numpy.nan)
+    m=numpy.full((e.shape[0]+1,e.shape[1]+1),numpy.nan)
+    m[1:,1:]=numpy.nanmax(e,axis=2).transpose()
+    p=numpy.zeros(m.shape)
+    p[1:,1:]=numpy.nanargmax(e,axis=2).transpose()
+    for i in range(1,m.shape[0]):
+      m[i,0],m[i,i],p[i,0]=m[i,i],numpy.nan,p[i,i]
+    h=ufal.chu_liu_edmonds.chu_liu_edmonds(m)[0]
+    if [0 for i in h if i==0]!=[0]:
+      m[:,0]+=numpy.where(m[:,0]==numpy.nanmax(m[[i for i,j in enumerate(h) if j==0],0]),0,numpy.nan)
+      m[[i for i,j in enumerate(h) if j==0]]+=[0 if i==0 or j==0 else numpy.nan for i,j in enumerate(h)]
+      h=ufal.chu_liu_edmonds.chu_liu_edmonds(m)[0]
+    u="# text = "+text+"\n"
+    v=[(s,e) for s,e in w["offset_mapping"] if s<e]
+    for i,(s,e) in enumerate(v,1):
+      q=self.model.config.id2label[p[i,h[i]]].split("|")
+      u+="\t".join([str(i),text[s:e],"_",q[0],"_","|".join(q[1:-1]),str(h[i]),q[-1],"_","_" if i<len(v) and e<v[i][0] else "SpaceAfter=No"])+"\n"
+    return u+"\n"
+
+nlp=UDgoeswith("KoichiYasuoka/roberta-base-thai-syllable-ud-goeswith")
+print(nlp("หลายหัวดีกว่าหัวเดียว"))
+```
+
+with [ufal.chu-liu-edmonds](https://pypi.org/project/ufal.chu-liu-edmonds/).
+Or without ufal.chu-liu-edmonds:
+
+```
+from transformers import pipeline
+nlp=pipeline("universal-dependencies","KoichiYasuoka/roberta-base-thai-syllable-ud-goeswith",trust_remote_code=True,aggregation_strategy="simple")
+print(nlp("หลายหัวดีกว่าหัวเดียว"))
+```
+

config.json

@@ -0,0 +1,517 @@
+{
+  "architectures": [
+    "RobertaForTokenClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "custom_pipelines": {
+    "universal-dependencies": {
+      "impl": "ud.UniversalDependenciesPipeline"
+    }
+  },
+  "eos_token_id": 2,
+  "gradient_checkpointing": false,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "-|_|dep",
+    "1": "ADP|_|acl",
+    "2": "ADP|_|advcl",
+    "3": "ADP|_|advmod",
+    "4": "ADP|_|appos",
+    "5": "ADP|_|case",
+    "6": "ADP|_|cc",
+    "7": "ADP|_|cc:preconj",
+    "8": "ADP|_|csubj",
+    "9": "ADP|_|fixed",
+    "10": "ADP|_|mark",
+    "11": "ADP|_|obl",
+    "12": "ADP|_|root",
+    "13": "ADV|PronType=Int|advmod",
+    "14": "ADV|_|advcl",
+    "15": "ADV|_|advmod",
+    "16": "ADV|_|aux",
+    "17": "ADV|_|cc",
+    "18": "ADV|_|ccomp",
+    "19": "ADV|_|conj",
+    "20": "ADV|_|fixed",
+    "21": "ADV|_|mark",
+    "22": "ADV|_|obj",
+    "23": "ADV|_|root",
+    "24": "ADV|_|xcomp",
+    "25": "AUX|_|advmod",
+    "26": "AUX|_|aux",
+    "27": "AUX|_|aux:pass",
+    "28": "AUX|_|ccomp",
+    "29": "AUX|_|conj",
+    "30": "AUX|_|cop",
+    "31": "AUX|_|mark",
+    "32": "CCONJ|_|advmod",
+    "33": "CCONJ|_|case",
+    "34": "CCONJ|_|cc",
+    "35": "CCONJ|_|compound",
+    "36": "CCONJ|_|conj",
+    "37": "CCONJ|_|fixed",
+    "38": "CCONJ|_|mark",
+    "39": "CCONJ|_|nsubj",
+    "40": "CCONJ|_|obl",
+    "41": "CCONJ|_|root",
+    "42": "DET|PronType=Int|det",
+    "43": "DET|_|advmod",
+    "44": "DET|_|case",
+    "45": "DET|_|cc:preconj",
+    "46": "DET|_|conj",
+    "47": "DET|_|det",
+    "48": "DET|_|det:predet",
+    "49": "DET|_|fixed",
+    "50": "DET|_|mark",
+    "51": "DET|_|nsubj",
+    "52": "DET|_|nsubj:pass",
+    "53": "DET|_|obj",
+    "54": "DET|_|obl",
+    "55": "DET|_|obl:tmod",
+    "56": "DET|_|root",
+    "57": "INTJ|_|acl",
+    "58": "INTJ|_|nsubj",
+    "59": "INTJ|_|root",
+    "60": "NOUN|_|acl",
+    "61": "NOUN|_|acl:relcl",
+    "62": "NOUN|_|advcl",
+    "63": "NOUN|_|advmod",
+    "64": "NOUN|_|appos",
+    "65": "NOUN|_|aux",
+    "66": "NOUN|_|case",
+    "67": "NOUN|_|cc",
+    "68": "NOUN|_|ccomp",
+    "69": "NOUN|_|clf",
+    "70": "NOUN|_|compound",
+    "71": "NOUN|_|conj",
+    "72": "NOUN|_|dislocated",
+    "73": "NOUN|_|fixed",
+    "74": "NOUN|_|flat:name",
+    "75": "NOUN|_|iobj",
+    "76": "NOUN|_|mark",
+    "77": "NOUN|_|nmod",
+    "78": "NOUN|_|nmod:poss",
+    "79": "NOUN|_|nsubj",
+    "80": "NOUN|_|nsubj:pass",
+    "81": "NOUN|_|obj",
+    "82": "NOUN|_|obl",
+    "83": "NOUN|_|obl:poss",
+    "84": "NOUN|_|obl:tmod",
+    "85": "NOUN|_|parataxis",
+    "86": "NOUN|_|root",
+    "87": "NOUN|_|vocative",
+    "88": "NOUN|_|xcomp",
+    "89": "NUM|_|acl",
+    "90": "NUM|_|acl:relcl",
+    "91": "NUM|_|advmod",
+    "92": "NUM|_|appos",
+    "93": "NUM|_|ccomp",
+    "94": "NUM|_|clf",
+    "95": "NUM|_|conj",
+    "96": "NUM|_|flat:name",
+    "97": "NUM|_|nmod",
+    "98": "NUM|_|nsubj",
+    "99": "NUM|_|nummod",
+    "100": "NUM|_|obj",
+    "101": "NUM|_|obl",
+    "102": "NUM|_|obl:poss",
+    "103": "NUM|_|obl:tmod",
+    "104": "NUM|_|root",
+    "105": "NUM|_|xcomp",
+    "106": "PART|Aspect=Perf|aux",
+    "107": "PART|Aspect=Perf|xcomp",
+    "108": "PART|Aspect=Prog|aux",
+    "109": "PART|Polarity=Neg|advmod",
+    "110": "PART|PronType=Int|acl",
+    "111": "PART|PronType=Int|advmod",
+    "112": "PART|PronType=Int|discourse",
+    "113": "PART|PronType=Int|obj",
+    "114": "PART|PronType=Int|root",
+    "115": "PART|_|acl",
+    "116": "PART|_|advmod",
+    "117": "PART|_|aux",
+    "118": "PART|_|cc",
+    "119": "PART|_|cc:preconj",
+    "120": "PART|_|ccomp",
+    "121": "PART|_|clf",
+    "122": "PART|_|compound",
+    "123": "PART|_|compound:prt",
+    "124": "PART|_|conj",
+    "125": "PART|_|discourse",
+    "126": "PART|_|fixed",
+    "127": "PART|_|mark",
+    "128": "PART|_|nmod",
+    "129": "PART|_|nmod:poss",
+    "130": "PART|_|nsubj",
+    "131": "PART|_|obj",
+    "132": "PART|_|obl",
+    "133": "PART|_|root",
+    "134": "PRON|Person=1|compound",
+    "135": "PRON|Person=1|nmod:poss",
+    "136": "PRON|Person=1|nsubj",
+    "137": "PRON|Person=1|nsubj:pass",
+    "138": "PRON|Person=1|obj",
+    "139": "PRON|Person=1|obl",
+    "140": "PRON|Person=1|obl:poss",
+    "141": "PRON|Person=2|compound",
+    "142": "PRON|Person=2|nmod:poss",
+    "143": "PRON|Person=2|nsubj",
+    "144": "PRON|Person=2|obj",
+    "145": "PRON|Person=2|obl",
+    "146": "PRON|Person=3|advmod",
+    "147": "PRON|Person=3|appos",
+    "148": "PRON|Person=3|compound",
+    "149": "PRON|Person=3|conj",
+    "150": "PRON|Person=3|nmod",
+    "151": "PRON|Person=3|nmod:poss",
+    "152": "PRON|Person=3|nsubj",
+    "153": "PRON|Person=3|nsubj:pass",
+    "154": "PRON|Person=3|obj",
+    "155": "PRON|Person=3|obl",
+    "156": "PRON|Person=3|obl:poss",
+    "157": "PRON|Person=3|reparandum",
+    "158": "PRON|Person=3|xcomp",
+    "159": "PRON|PronType=Int|nsubj",
+    "160": "PRON|PronType=Int|obj",
+    "161": "PRON|PronType=Int|obl",
+    "162": "PRON|PronType=Int|root",
+    "163": "PRON|_|acl",
+    "164": "PRON|_|acl:relcl",
+    "165": "PRON|_|advcl",
+    "166": "PRON|_|advmod",
+    "167": "PRON|_|ccomp",
+    "168": "PRON|_|compound",
+    "169": "PRON|_|conj",
+    "170": "PRON|_|fixed",
+    "171": "PRON|_|nmod",
+    "172": "PRON|_|nmod:poss",
+    "173": "PRON|_|nsubj",
+    "174": "PRON|_|obj",
+    "175": "PRON|_|obl",
+    "176": "PRON|_|obl:poss",
+    "177": "PRON|_|root",
+    "178": "PRON|_|xcomp",
+    "179": "PROPN|_|acl",
+    "180": "PROPN|_|acl:relcl",
+    "181": "PROPN|_|advmod",
+    "182": "PROPN|_|appos",
+    "183": "PROPN|_|aux",
+    "184": "PROPN|_|cc",
+    "185": "PROPN|_|ccomp",
+    "186": "PROPN|_|clf",
+    "187": "PROPN|_|compound",
+    "188": "PROPN|_|conj",
+    "189": "PROPN|_|flat:name",
+    "190": "PROPN|_|goeswith",
+    "191": "PROPN|_|nmod",
+    "192": "PROPN|_|nmod:poss",
+    "193": "PROPN|_|nsubj",
+    "194": "PROPN|_|nsubj:pass",
+    "195": "PROPN|_|obj",
+    "196": "PROPN|_|obl",
+    "197": "PROPN|_|obl:poss",
+    "198": "PROPN|_|obl:tmod",
+    "199": "PROPN|_|root",
+    "200": "PROPN|_|xcomp",
+    "201": "PUNCT|_|advmod",
+    "202": "PUNCT|_|clf",
+    "203": "PUNCT|_|punct",
+    "204": "PUNCT|_|root",
+    "205": "SCONJ|_|mark",
+    "206": "SYM|_|advmod",
+    "207": "SYM|_|clf",
+    "208": "SYM|_|nsubj",
+    "209": "SYM|_|obj",
+    "210": "SYM|_|obl",
+    "211": "VERB|Mood=Imp|aux",
+    "212": "VERB|Mood=Imp|xcomp",
+    "213": "VERB|Voice=Pass|aux",
+    "214": "VERB|Voice=Pass|aux:pass",
+    "215": "VERB|_|acl",
+    "216": "VERB|_|acl:relcl",
+    "217": "VERB|_|advcl",
+    "218": "VERB|_|advmod",
+    "219": "VERB|_|appos",
+    "220": "VERB|_|aux",
+    "221": "VERB|_|case",
+    "222": "VERB|_|cc",
+    "223": "VERB|_|ccomp",
+    "224": "VERB|_|compound",
+    "225": "VERB|_|conj",
+    "226": "VERB|_|csubj",
+    "227": "VERB|_|fixed",
+    "228": "VERB|_|mark",
+    "229": "VERB|_|nmod",
+    "230": "VERB|_|nmod:poss",
+    "231": "VERB|_|nsubj",
+    "232": "VERB|_|obj",
+    "233": "VERB|_|obl",
+    "234": "VERB|_|obl:poss",
+    "235": "VERB|_|parataxis",
+    "236": "VERB|_|root",
+    "237": "VERB|_|xcomp",
+    "238": "X|_|goeswith"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "-|_|dep": 0,
+    "ADP|_|acl": 1,
+    "ADP|_|advcl": 2,
+    "ADP|_|advmod": 3,
+    "ADP|_|appos": 4,
+    "ADP|_|case": 5,
+    "ADP|_|cc": 6,
+    "ADP|_|cc:preconj": 7,
+    "ADP|_|csubj": 8,
+    "ADP|_|fixed": 9,
+    "ADP|_|mark": 10,
+    "ADP|_|obl": 11,
+    "ADP|_|root": 12,
+    "ADV|PronType=Int|advmod": 13,
+    "ADV|_|advcl": 14,
+    "ADV|_|advmod": 15,
+    "ADV|_|aux": 16,
+    "ADV|_|cc": 17,
+    "ADV|_|ccomp": 18,
+    "ADV|_|conj": 19,
+    "ADV|_|fixed": 20,
+    "ADV|_|mark": 21,
+    "ADV|_|obj": 22,
+    "ADV|_|root": 23,
+    "ADV|_|xcomp": 24,
+    "AUX|_|advmod": 25,
+    "AUX|_|aux": 26,
+    "AUX|_|aux:pass": 27,
+    "AUX|_|ccomp": 28,
+    "AUX|_|conj": 29,
+    "AUX|_|cop": 30,
+    "AUX|_|mark": 31,
+    "CCONJ|_|advmod": 32,
+    "CCONJ|_|case": 33,
+    "CCONJ|_|cc": 34,
+    "CCONJ|_|compound": 35,
+    "CCONJ|_|conj": 36,
+    "CCONJ|_|fixed": 37,
+    "CCONJ|_|mark": 38,
+    "CCONJ|_|nsubj": 39,
+    "CCONJ|_|obl": 40,
+    "CCONJ|_|root": 41,
+    "DET|PronType=Int|det": 42,
+    "DET|_|advmod": 43,
+    "DET|_|case": 44,
+    "DET|_|cc:preconj": 45,
+    "DET|_|conj": 46,
+    "DET|_|det": 47,
+    "DET|_|det:predet": 48,
+    "DET|_|fixed": 49,
+    "DET|_|mark": 50,
+    "DET|_|nsubj": 51,
+    "DET|_|nsubj:pass": 52,
+    "DET|_|obj": 53,
+    "DET|_|obl": 54,
+    "DET|_|obl:tmod": 55,
+    "DET|_|root": 56,
+    "INTJ|_|acl": 57,
+    "INTJ|_|nsubj": 58,
+    "INTJ|_|root": 59,
+    "NOUN|_|acl": 60,
+    "NOUN|_|acl:relcl": 61,
+    "NOUN|_|advcl": 62,
+    "NOUN|_|advmod": 63,
+    "NOUN|_|appos": 64,
+    "NOUN|_|aux": 65,
+    "NOUN|_|case": 66,
+    "NOUN|_|cc": 67,
+    "NOUN|_|ccomp": 68,
+    "NOUN|_|clf": 69,
+    "NOUN|_|compound": 70,
+    "NOUN|_|conj": 71,
+    "NOUN|_|dislocated": 72,
+    "NOUN|_|fixed": 73,
+    "NOUN|_|flat:name": 74,
+    "NOUN|_|iobj": 75,
+    "NOUN|_|mark": 76,
+    "NOUN|_|nmod": 77,
+    "NOUN|_|nmod:poss": 78,
+    "NOUN|_|nsubj": 79,
+    "NOUN|_|nsubj:pass": 80,
+    "NOUN|_|obj": 81,
+    "NOUN|_|obl": 82,
+    "NOUN|_|obl:poss": 83,
+    "NOUN|_|obl:tmod": 84,
+    "NOUN|_|parataxis": 85,
+    "NOUN|_|root": 86,
+    "NOUN|_|vocative": 87,
+    "NOUN|_|xcomp": 88,
+    "NUM|_|acl": 89,
+    "NUM|_|acl:relcl": 90,
+    "NUM|_|advmod": 91,
+    "NUM|_|appos": 92,
+    "NUM|_|ccomp": 93,
+    "NUM|_|clf": 94,
+    "NUM|_|conj": 95,
+    "NUM|_|flat:name": 96,
+    "NUM|_|nmod": 97,
+    "NUM|_|nsubj": 98,
+    "NUM|_|nummod": 99,
+    "NUM|_|obj": 100,
+    "NUM|_|obl": 101,
+    "NUM|_|obl:poss": 102,
+    "NUM|_|obl:tmod": 103,
+    "NUM|_|root": 104,
+    "NUM|_|xcomp": 105,
+    "PART|Aspect=Perf|aux": 106,
+    "PART|Aspect=Perf|xcomp": 107,
+    "PART|Aspect=Prog|aux": 108,
+    "PART|Polarity=Neg|advmod": 109,
+    "PART|PronType=Int|acl": 110,
+    "PART|PronType=Int|advmod": 111,
+    "PART|PronType=Int|discourse": 112,
+    "PART|PronType=Int|obj": 113,
+    "PART|PronType=Int|root": 114,
+    "PART|_|acl": 115,
+    "PART|_|advmod": 116,
+    "PART|_|aux": 117,
+    "PART|_|cc": 118,
+    "PART|_|cc:preconj": 119,
+    "PART|_|ccomp": 120,
+    "PART|_|clf": 121,
+    "PART|_|compound": 122,
+    "PART|_|compound:prt": 123,
+    "PART|_|conj": 124,
+    "PART|_|discourse": 125,
+    "PART|_|fixed": 126,
+    "PART|_|mark": 127,
+    "PART|_|nmod": 128,
+    "PART|_|nmod:poss": 129,
+    "PART|_|nsubj": 130,
+    "PART|_|obj": 131,
+    "PART|_|obl": 132,
+    "PART|_|root": 133,
+    "PRON|Person=1|compound": 134,
+    "PRON|Person=1|nmod:poss": 135,
+    "PRON|Person=1|nsubj": 136,
+    "PRON|Person=1|nsubj:pass": 137,
+    "PRON|Person=1|obj": 138,
+    "PRON|Person=1|obl": 139,
+    "PRON|Person=1|obl:poss": 140,
+    "PRON|Person=2|compound": 141,
+    "PRON|Person=2|nmod:poss": 142,
+    "PRON|Person=2|nsubj": 143,
+    "PRON|Person=2|obj": 144,
+    "PRON|Person=2|obl": 145,
+    "PRON|Person=3|advmod": 146,
+    "PRON|Person=3|appos": 147,
+    "PRON|Person=3|compound": 148,
+    "PRON|Person=3|conj": 149,
+    "PRON|Person=3|nmod": 150,
+    "PRON|Person=3|nmod:poss": 151,
+    "PRON|Person=3|nsubj": 152,
+    "PRON|Person=3|nsubj:pass": 153,
+    "PRON|Person=3|obj": 154,
+    "PRON|Person=3|obl": 155,
+    "PRON|Person=3|obl:poss": 156,
+    "PRON|Person=3|reparandum": 157,
+    "PRON|Person=3|xcomp": 158,
+    "PRON|PronType=Int|nsubj": 159,
+    "PRON|PronType=Int|obj": 160,
+    "PRON|PronType=Int|obl": 161,
+    "PRON|PronType=Int|root": 162,
+    "PRON|_|acl": 163,
+    "PRON|_|acl:relcl": 164,
+    "PRON|_|advcl": 165,
+    "PRON|_|advmod": 166,
+    "PRON|_|ccomp": 167,
+    "PRON|_|compound": 168,
+    "PRON|_|conj": 169,
+    "PRON|_|fixed": 170,
+    "PRON|_|nmod": 171,
+    "PRON|_|nmod:poss": 172,
+    "PRON|_|nsubj": 173,
+    "PRON|_|obj": 174,
+    "PRON|_|obl": 175,
+    "PRON|_|obl:poss": 176,
+    "PRON|_|root": 177,
+    "PRON|_|xcomp": 178,
+    "PROPN|_|acl": 179,
+    "PROPN|_|acl:relcl": 180,
+    "PROPN|_|advmod": 181,
+    "PROPN|_|appos": 182,
+    "PROPN|_|aux": 183,
+    "PROPN|_|cc": 184,
+    "PROPN|_|ccomp": 185,
+    "PROPN|_|clf": 186,
+    "PROPN|_|compound": 187,
+    "PROPN|_|conj": 188,
+    "PROPN|_|flat:name": 189,
+    "PROPN|_|goeswith": 190,
+    "PROPN|_|nmod": 191,
+    "PROPN|_|nmod:poss": 192,
+    "PROPN|_|nsubj": 193,
+    "PROPN|_|nsubj:pass": 194,
+    "PROPN|_|obj": 195,
+    "PROPN|_|obl": 196,
+    "PROPN|_|obl:poss": 197,
+    "PROPN|_|obl:tmod": 198,
+    "PROPN|_|root": 199,
+    "PROPN|_|xcomp": 200,
+    "PUNCT|_|advmod": 201,
+    "PUNCT|_|clf": 202,
+    "PUNCT|_|punct": 203,
+    "PUNCT|_|root": 204,
+    "SCONJ|_|mark": 205,
+    "SYM|_|advmod": 206,
+    "SYM|_|clf": 207,
+    "SYM|_|nsubj": 208,
+    "SYM|_|obj": 209,
+    "SYM|_|obl": 210,
+    "VERB|Mood=Imp|aux": 211,
+    "VERB|Mood=Imp|xcomp": 212,
+    "VERB|Voice=Pass|aux": 213,
+    "VERB|Voice=Pass|aux:pass": 214,
+    "VERB|_|acl": 215,
+    "VERB|_|acl:relcl": 216,
+    "VERB|_|advcl": 217,
+    "VERB|_|advmod": 218,
+    "VERB|_|appos": 219,
+    "VERB|_|aux": 220,
+    "VERB|_|case": 221,
+    "VERB|_|cc": 222,
+    "VERB|_|ccomp": 223,
+    "VERB|_|compound": 224,
+    "VERB|_|conj": 225,
+    "VERB|_|csubj": 226,
+    "VERB|_|fixed": 227,
+    "VERB|_|mark": 228,
+    "VERB|_|nmod": 229,
+    "VERB|_|nmod:poss": 230,
+    "VERB|_|nsubj": 231,
+    "VERB|_|obj": 232,
+    "VERB|_|obl": 233,
+    "VERB|_|obl:poss": 234,
+    "VERB|_|parataxis": 235,
+    "VERB|_|root": 236,
+    "VERB|_|xcomp": 237,
+    "X|_|goeswith": 238
+  },
+  "layer_norm_eps": 1e-05,
+  "mask_token_id": 4,
+  "max_position_embeddings": 514,
+  "model_type": "roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "tokenizer_class": "BertTokenizerFast",
+  "torch_dtype": "float32",
+  "transformers_version": "4.22.1",
+  "type_vocab_size": 1,
+  "unk_token_id": 3,
+  "use_cache": true,
+  "vocab_size": 16582
+}

maker.py

@@ -0,0 +1,55 @@
+#! /usr/bin/python3
+src="KoichiYasuoka/roberta-base-thai-syllable"
+tgt="KoichiYasuoka/roberta-base-thai-syllable-ud-goeswith"
+url="https://github.com/KoichiYasuoka/spaCy-Thai"
+import os
+d=os.path.join(os.path.basename(url),"UD_Thai-Corpora")
+os.system("test -d {} || git clone --depth=1 {}".format(d,url))
+s='{if(NF>0)u=u$0"\\n";else{f=FILENAME;if(u~/\\t0\\troot\\t/)print u>(f~/-dev/?"dev":f~/-test/?"test":"train")".conllu";u=""}}'
+os.system("nawk -F'\\t' '{}' {}/*-ud-*.conllu".format(s,d))
+class UDgoeswithDataset(object):
+  def __init__(self,conllu,tokenizer):
+    self.ids,self.tags,label=[],[],set()
+    with open(conllu,"r",encoding="utf-8") as r:
+      cls,sep,msk=tokenizer.cls_token_id,tokenizer.sep_token_id,tokenizer.mask_token_id
+      dep,c="-|_|dep",[]
+      for s in r:
+        t=s.split("\t")
+        if len(t)==10 and t[0].isdecimal():
+          c.append(t)
+        elif c!=[]:
+          v=tokenizer([t[1] for t in c],add_special_tokens=False)["input_ids"]
+          for i in range(len(v)-1,-1,-1):
+            for j in range(1,len(v[i])):
+              c.insert(i+1,[c[i][0],"_","_","X","_","_",c[i][0],"goeswith","_","_"])
+          y=["0"]+[t[0] for t in c]
+          h=[i if t[6]=="0" else y.index(t[6]) for i,t in enumerate(c,1)]
+          p,v=[t[3]+"|"+t[5]+"|"+t[7] for t in c],sum(v,[])
+          self.ids.append([cls]+v+[sep])
+          self.tags.append([dep]+p+[dep])
+          label=set(sum([self.tags[-1],list(label)],[]))
+          for i,k in enumerate(v):
+            self.ids.append([cls]+v[0:i]+[msk]+v[i+1:]+[sep,k])
+            self.tags.append([dep]+[t if h[j]==i+1 else dep for j,t in enumerate(p)]+[dep,dep])
+          c=[]
+    self.label2id={l:i for i,l in enumerate(sorted(label))}
+  def __call__(*args):
+    label=set(sum([list(t.label2id) for t in args],[]))
+    lid={l:i for i,l in enumerate(sorted(label))}
+    for t in args:
+      t.label2id=lid
+    return lid
+  __len__=lambda self:len(self.ids)
+  __getitem__=lambda self,i:{"input_ids":self.ids[i],"labels":[self.label2id[t] for t in self.tags[i]]}
+from transformers import AutoTokenizer,AutoConfig,AutoModelForTokenClassification,DataCollatorForTokenClassification,TrainingArguments,Trainer
+tkz=AutoTokenizer.from_pretrained(src)
+trainDS=UDgoeswithDataset("train.conllu",tkz)
+devDS=UDgoeswithDataset("dev.conllu",tkz)
+testDS=UDgoeswithDataset("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()})
+arg=TrainingArguments(num_train_epochs=3,per_device_train_batch_size=32,output_dir="/tmp",overwrite_output_dir=True,save_total_limit=2,evaluation_strategy="epoch",learning_rate=5e-05,warmup_ratio=0.1)
+trn=Trainer(args=arg,data_collator=DataCollatorForTokenClassification(tkz),model=AutoModelForTokenClassification.from_pretrained(src,config=cfg),train_dataset=trainDS,eval_dataset=devDS)
+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 sha256:5614b6b4e9336d394455eb9e54aabddea48d6c140c8b665e0f8961d0d5e65dfc
+size 393552177

special_tokens_map.json

@@ -0,0 +1,7 @@
+{
+  "cls_token": "<s>",
+  "mask_token": "<mask>",
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "unk_token": "<unk>"
+}

tokenizer_config.json

@@ -0,0 +1,14 @@
+{
+  "cls_token": "<s>",
+  "do_basic_tokenize": true,
+  "do_lower_case": false,
+  "mask_token": "<mask>",
+  "model_max_length": 416,
+  "never_split": null,
+  "pad_token": "<pad>",
+  "sep_token": "</s>",
+  "strip_accents": false,
+  "tokenize_chinese_chars": true,
+  "tokenizer_class": "BertTokenizerFast",
+  "unk_token": "<unk>"
+}

ud.py

@@ -0,0 +1,61 @@
+from transformers import TokenClassificationPipeline
+
+class UniversalDependenciesPipeline(TokenClassificationPipeline):
+  def _forward(self,model_input):
+    import torch
+    v=model_input["input_ids"][0].tolist()
+    with torch.no_grad():
+      e=self.model(input_ids=torch.tensor([v[0:i]+[self.tokenizer.mask_token_id]+v[i+1:]+[j] for i,j in enumerate(v[1:-1],1)]))
+    return {"logits":e.logits[:,1:-2,:],**model_input}
+  def postprocess(self,model_output,**kwargs):
+    import numpy
+    e=model_output["logits"].numpy()
+    r=[1 if i==0 else -1 if j.endswith("|root") else 0 for i,j in sorted(self.model.config.id2label.items())]
+    e+=numpy.where(numpy.add.outer(numpy.identity(e.shape[0]),r)==0,0,numpy.nan)
+    g=self.model.config.label2id["X|_|goeswith"]
+    r=numpy.tri(e.shape[0])
+    for i in range(e.shape[0]):
+      for j in range(i+2,e.shape[1]):
+        r[i,j]=r[i,j-1] if numpy.nanargmax(e[i,j-1])==g else 1
+    e[:,:,g]+=numpy.where(r==0,0,numpy.nan)
+    m,p=numpy.nanmax(e,axis=2),numpy.nanargmax(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.nanargmax(m[z,z])],numpy.nanmin(m)-numpy.nanmax(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)
+    v=[(s,e) for s,e in model_output["offset_mapping"][0].tolist() if s<e]
+    q=[self.model.config.id2label[p[j,i]].split("|") for i,j in enumerate(h)]
+    g="aggregation_strategy" in kwargs and kwargs["aggregation_strategy"]!="none"
+    if g:
+      for i,j in reversed(list(enumerate(q[1:],1))):
+        if j[-1]=="goeswith" and set([t[-1] for t in q[h[i]+1:i+1]])=={"goeswith"}:
+          h=[b if i>b else b-1 for a,b in enumerate(h) if i!=a]
+          v[i-1]=(v[i-1][0],v.pop(i)[1])
+          q.pop(i)
+    t=model_output["sentence"].replace("\n"," ")
+    u="# text = "+t+"\n"
+    for i,(s,e) in enumerate(v):
+      u+="\t".join([str(i+1),t[s:e],t[s:e] if g else "_",q[i][0],"_","|".join(q[i][1:-1]),str(0 if h[i]==i else h[i]+1),q[i][-1],"_","_" if i+1<len(v) and e<v[i+1][0] else "SpaceAfter=No"])+"\n"
+    return u+"\n"
+  def chu_liu_edmonds(self,matrix):
+    import numpy
+    h=numpy.nanargmax(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.nanmax(matrix,axis=0)
+    m=numpy.block([[z[x,:][:,x],numpy.nanmax(z[x,:][:,y],axis=1).reshape(len(x),1)],[numpy.nanmax(z[y,:][:,x],axis=0),numpy.nanmax(z[y,y])]])
+    k=[j if i==len(x) else x[j] if j<len(x) else y[numpy.nanargmax(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.nanargmax(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