initial release

README.md

@@ -0,0 +1,98 @@
+---
+language:
+- "th"
+tags:
+- "thai"
+- "question-answering"
+- "dependency-parsing"
+datasets:
+- "universal_dependencies"
+license: "apache-2.0"
+pipeline_tag: "question-answering"
+widget:
+- text: "กว่า"
+  context: "หลายหัวดีกว่าหัวเดียว"
+- text: "หลาย"
+  context: "หลายหัวดีกว่าหัวเดียว"
+- text: "หัว"
+  context: "หลาย[MASK]ดีกว่าหัวเดียว"
+---
+
+# roberta-base-thai-spm-ud-head
+
+## Model Description
+
+This is a DeBERTa(V2) model pretrained on Thai Wikipedia texts for dependency-parsing (head-detection on Universal Dependencies) as question-answering, derived from [roberta-base-thai-spm](https://huggingface.co/KoichiYasuoka/roberta-base-thai-spm). Use [MASK] inside `context` to avoid ambiguity when specifying a multiple-used word as `question`.
+
+## How to Use
+
+```py
+import torch
+from transformers import AutoTokenizer,AutoModelForQuestionAnswering
+tokenizer=AutoTokenizer.from_pretrained("KoichiYasuoka/roberta-base-thai-spm-ud-head")
+model=AutoModelForQuestionAnswering.from_pretrained("KoichiYasuoka/roberta-base-thai-spm-ud-head")
+question="กว่า"
+context="หลายหัวดีกว่าหัวเดียว"
+inputs=tokenizer(question,context,return_tensors="pt",return_offsets_mapping=True)
+offsets=inputs.pop("offset_mapping").tolist()[0]
+outputs=model(**inputs)
+start,end=torch.argmax(outputs.start_logits),torch.argmax(outputs.end_logits)
+print(context[offsets[start][0]:offsets[end][-1]])
+```
+
+or (with [ufal.chu-liu-edmonds](https://pypi.org/project/ufal.chu-liu-edmonds/))
+
+```py
+class TransformersUD(object):
+  def __init__(self,bert):
+    import os
+    from transformers import (AutoTokenizer,AutoModelForQuestionAnswering,
+      AutoModelForTokenClassification,AutoConfig,TokenClassificationPipeline)
+    self.tokenizer=AutoTokenizer.from_pretrained(bert)
+    self.model=AutoModelForQuestionAnswering.from_pretrained(bert)
+    x=AutoModelForTokenClassification.from_pretrained
+    if os.path.isdir(bert):
+      d,t=x(os.path.join(bert,"deprel")),x(os.path.join(bert,"tagger"))
+    else:
+      from transformers.file_utils import hf_bucket_url
+      c=AutoConfig.from_pretrained(hf_bucket_url(bert,"deprel/config.json"))
+      d=x(hf_bucket_url(bert,"deprel/pytorch_model.bin"),config=c)
+      s=AutoConfig.from_pretrained(hf_bucket_url(bert,"tagger/config.json"))
+      t=x(hf_bucket_url(bert,"tagger/pytorch_model.bin"),config=s)
+    self.deprel=TokenClassificationPipeline(model=d,tokenizer=self.tokenizer,
+      aggregation_strategy="simple")
+    self.tagger=TokenClassificationPipeline(model=t,tokenizer=self.tokenizer)
+  def __call__(self,text):
+    import numpy,torch,ufal.chu_liu_edmonds
+    w=[(t["start"],t["end"],t["entity_group"]) for t in self.deprel(text)]
+    z,n={t["start"]:t["entity"].split("|") for t in self.tagger(text)},len(w)
+    r,m=[text[s:e] for s,e,p in w],numpy.full((n+1,n+1),numpy.nan)
+    v,c=self.tokenizer(r,add_special_tokens=False)["input_ids"],[]
+    for i,t in enumerate(v):
+      q=[self.tokenizer.cls_token_id]+t+[self.tokenizer.sep_token_id]
+      c.append([q]+v[0:i]+[[self.tokenizer.mask_token_id]]+v[i+1:]+[[q[-1]]])
+    b=[[len(sum(x[0:j+1],[])) for j in range(len(x))] for x in c]
+    with torch.no_grad():
+      d=self.model(input_ids=torch.tensor([sum(x,[]) for x in c]),
+        token_type_ids=torch.tensor([[0]*x[0]+[1]*(x[-1]-x[0]) for x in b]))
+    s,e=d.start_logits.tolist(),d.end_logits.tolist()
+    for i in range(n):
+      for j in range(n):
+        m[i+1,0 if i==j else j+1]=s[i][b[i][j]]+e[i][b[i][j+1]-1]
+    h=ufal.chu_liu_edmonds.chu_liu_edmonds(m)[0]
+    if [0 for i in h if i==0]!=[0]:
+      i=([p for s,e,p in w]+["root"]).index("root")
+      j=i+1 if i<n else numpy.nanargmax(m[:,0])
+      m[0:j,0]=m[j+1:,0]=numpy.nan
+      h=ufal.chu_liu_edmonds.chu_liu_edmonds(m)[0]
+    u="# text = "+text.replace("\n"," ")+"\n"
+    for i,(s,e,p) in enumerate(w,1):
+      p="root" if h[i]==0 else "dep" if p=="root" else p
+      u+="\t".join([str(i),r[i-1],"_",z[s][0][2:],"_","|".join(z[s][1:]),
+        str(h[i]),p,"_","_" if i<n and w[i][0]<e else "SpaceAfter=No"])+"\n"
+    return u+"\n"
+
+nlp=TransformersUD("KoichiYasuoka/roberta-base-thai-spm-ud-head")
+print(nlp("หลายหัวดีกว่าหัวเดียว"))
+```
+

config.json

@@ -0,0 +1,27 @@
+{
+  "architectures": [
+    "RobertaForQuestionAnswering"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "tokenizer_class": "RemBertTokenizerFast",
+  "torch_dtype": "float32",
+  "transformers_version": "4.19.4",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 3005
+}

deprel/config.json

@@ -0,0 +1,187 @@
+{
+  "architectures": [
+    "RobertaForTokenClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "B-acl",
+    "1": "B-acl:relcl",
+    "2": "B-advcl",
+    "3": "B-advmod",
+    "4": "B-appos",
+    "5": "B-aux",
+    "6": "B-aux:pass",
+    "7": "B-case",
+    "8": "B-cc",
+    "9": "B-cc:preconj",
+    "10": "B-ccomp",
+    "11": "B-clf",
+    "12": "B-compound",
+    "13": "B-compound:prt",
+    "14": "B-conj",
+    "15": "B-cop",
+    "16": "B-csubj",
+    "17": "B-det",
+    "18": "B-det:predet",
+    "19": "B-discourse",
+    "20": "B-dislocated",
+    "21": "B-fixed",
+    "22": "B-flat:name",
+    "23": "B-goeswith",
+    "24": "B-iobj",
+    "25": "B-mark",
+    "26": "B-nmod",
+    "27": "B-nmod:poss",
+    "28": "B-nsubj",
+    "29": "B-nsubj:pass",
+    "30": "B-nummod",
+    "31": "B-obj",
+    "32": "B-obl",
+    "33": "B-obl:poss",
+    "34": "B-obl:tmod",
+    "35": "B-parataxis",
+    "36": "B-punct",
+    "37": "B-reparandum",
+    "38": "B-root",
+    "39": "B-vocative",
+    "40": "B-xcomp",
+    "41": "I-acl",
+    "42": "I-acl:relcl",
+    "43": "I-advcl",
+    "44": "I-advmod",
+    "45": "I-appos",
+    "46": "I-aux",
+    "47": "I-aux:pass",
+    "48": "I-case",
+    "49": "I-cc",
+    "50": "I-ccomp",
+    "51": "I-clf",
+    "52": "I-compound",
+    "53": "I-conj",
+    "54": "I-cop",
+    "55": "I-csubj",
+    "56": "I-det",
+    "57": "I-det:predet",
+    "58": "I-discourse",
+    "59": "I-dislocated",
+    "60": "I-fixed",
+    "61": "I-flat:name",
+    "62": "I-goeswith",
+    "63": "I-mark",
+    "64": "I-nmod",
+    "65": "I-nmod:poss",
+    "66": "I-nsubj",
+    "67": "I-nsubj:pass",
+    "68": "I-nummod",
+    "69": "I-obj",
+    "70": "I-obl",
+    "71": "I-obl:poss",
+    "72": "I-obl:tmod",
+    "73": "I-parataxis",
+    "74": "I-punct",
+    "75": "I-root",
+    "76": "I-vocative",
+    "77": "I-xcomp"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "B-acl": 0,
+    "B-acl:relcl": 1,
+    "B-advcl": 2,
+    "B-advmod": 3,
+    "B-appos": 4,
+    "B-aux": 5,
+    "B-aux:pass": 6,
+    "B-case": 7,
+    "B-cc": 8,
+    "B-cc:preconj": 9,
+    "B-ccomp": 10,
+    "B-clf": 11,
+    "B-compound": 12,
+    "B-compound:prt": 13,
+    "B-conj": 14,
+    "B-cop": 15,
+    "B-csubj": 16,
+    "B-det": 17,
+    "B-det:predet": 18,
+    "B-discourse": 19,
+    "B-dislocated": 20,
+    "B-fixed": 21,
+    "B-flat:name": 22,
+    "B-goeswith": 23,
+    "B-iobj": 24,
+    "B-mark": 25,
+    "B-nmod": 26,
+    "B-nmod:poss": 27,
+    "B-nsubj": 28,
+    "B-nsubj:pass": 29,
+    "B-nummod": 30,
+    "B-obj": 31,
+    "B-obl": 32,
+    "B-obl:poss": 33,
+    "B-obl:tmod": 34,
+    "B-parataxis": 35,
+    "B-punct": 36,
+    "B-reparandum": 37,
+    "B-root": 38,
+    "B-vocative": 39,
+    "B-xcomp": 40,
+    "I-acl": 41,
+    "I-acl:relcl": 42,
+    "I-advcl": 43,
+    "I-advmod": 44,
+    "I-appos": 45,
+    "I-aux": 46,
+    "I-aux:pass": 47,
+    "I-case": 48,
+    "I-cc": 49,
+    "I-ccomp": 50,
+    "I-clf": 51,
+    "I-compound": 52,
+    "I-conj": 53,
+    "I-cop": 54,
+    "I-csubj": 55,
+    "I-det": 56,
+    "I-det:predet": 57,
+    "I-discourse": 58,
+    "I-dislocated": 59,
+    "I-fixed": 60,
+    "I-flat:name": 61,
+    "I-goeswith": 62,
+    "I-mark": 63,
+    "I-nmod": 64,
+    "I-nmod:poss": 65,
+    "I-nsubj": 66,
+    "I-nsubj:pass": 67,
+    "I-nummod": 68,
+    "I-obj": 69,
+    "I-obl": 70,
+    "I-obl:poss": 71,
+    "I-obl:tmod": 72,
+    "I-parataxis": 73,
+    "I-punct": 74,
+    "I-root": 75,
+    "I-vocative": 76,
+    "I-xcomp": 77
+  },
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "tokenizer_class": "RemBertTokenizerFast",
+  "torch_dtype": "float32",
+  "transformers_version": "4.19.4",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 3005
+}

deprel/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5daa7d4b8c3c6e0ab27c91a0cdc7953972c61a981195e1d714ab33c46853838e
+size 351345265

deprel/sentencepiece.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b
+size 1

deprel/special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": {"content": "[MASK]", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true}}

deprel/tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": false, "remove_space": true, "keep_accents": true, "bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": {"content": "[MASK]", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "model_max_length": 512, "tokenizer_class": "RemBertTokenizerFast"}

pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f8e2ad36dfd34f137d576a17c0f358dd1dac8eaf003475bdfd61581e7902b079
+size 351111537

sentencepiece.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b
+size 1

special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": {"content": "[MASK]", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true}}

tagger/config.json

@@ -0,0 +1,133 @@
+{
+  "architectures": [
+    "RobertaForTokenClassification"
+  ],
+  "attention_probs_dropout_prob": 0.1,
+  "bos_token_id": 0,
+  "classifier_dropout": null,
+  "eos_token_id": 2,
+  "hidden_act": "gelu",
+  "hidden_dropout_prob": 0.1,
+  "hidden_size": 768,
+  "id2label": {
+    "0": "B-ADP|_",
+    "1": "B-ADV|PronType=Int",
+    "2": "B-ADV|_",
+    "3": "B-AUX|_",
+    "4": "B-CCONJ|_",
+    "5": "B-DET|PronType=Int",
+    "6": "B-DET|_",
+    "7": "B-INTJ|_",
+    "8": "B-NOUN|_",
+    "9": "B-NUM|_",
+    "10": "B-PART|Aspect=Perf",
+    "11": "B-PART|Aspect=Prog",
+    "12": "B-PART|Polarity=Neg",
+    "13": "B-PART|PronType=Int",
+    "14": "B-PART|_",
+    "15": "B-PRON|Person=1",
+    "16": "B-PRON|Person=2",
+    "17": "B-PRON|Person=3",
+    "18": "B-PRON|PronType=Int",
+    "19": "B-PRON|_",
+    "20": "B-PROPN|_",
+    "21": "B-PUNCT|_",
+    "22": "B-SCONJ|_",
+    "23": "B-SYM|_",
+    "24": "B-VERB|Mood=Imp",
+    "25": "B-VERB|Voice=Pass",
+    "26": "B-VERB|_",
+    "27": "B-X|_",
+    "28": "I-ADP|_",
+    "29": "I-ADV|PronType=Int",
+    "30": "I-ADV|_",
+    "31": "I-AUX|_",
+    "32": "I-CCONJ|_",
+    "33": "I-DET|_",
+    "34": "I-INTJ|_",
+    "35": "I-NOUN|_",
+    "36": "I-NUM|_",
+    "37": "I-PART|Aspect=Perf",
+    "38": "I-PART|Aspect=Prog",
+    "39": "I-PART|Polarity=Neg",
+    "40": "I-PART|PronType=Int",
+    "41": "I-PART|_",
+    "42": "I-PRON|Person=1",
+    "43": "I-PRON|Person=2",
+    "44": "I-PRON|Person=3",
+    "45": "I-PRON|_",
+    "46": "I-PROPN|_",
+    "47": "I-PUNCT|_",
+    "48": "I-VERB|Voice=Pass",
+    "49": "I-VERB|_",
+    "50": "I-X|_"
+  },
+  "initializer_range": 0.02,
+  "intermediate_size": 3072,
+  "label2id": {
+    "B-ADP|_": 0,
+    "B-ADV|PronType=Int": 1,
+    "B-ADV|_": 2,
+    "B-AUX|_": 3,
+    "B-CCONJ|_": 4,
+    "B-DET|PronType=Int": 5,
+    "B-DET|_": 6,
+    "B-INTJ|_": 7,
+    "B-NOUN|_": 8,
+    "B-NUM|_": 9,
+    "B-PART|Aspect=Perf": 10,
+    "B-PART|Aspect=Prog": 11,
+    "B-PART|Polarity=Neg": 12,
+    "B-PART|PronType=Int": 13,
+    "B-PART|_": 14,
+    "B-PRON|Person=1": 15,
+    "B-PRON|Person=2": 16,
+    "B-PRON|Person=3": 17,
+    "B-PRON|PronType=Int": 18,
+    "B-PRON|_": 19,
+    "B-PROPN|_": 20,
+    "B-PUNCT|_": 21,
+    "B-SCONJ|_": 22,
+    "B-SYM|_": 23,
+    "B-VERB|Mood=Imp": 24,
+    "B-VERB|Voice=Pass": 25,
+    "B-VERB|_": 26,
+    "B-X|_": 27,
+    "I-ADP|_": 28,
+    "I-ADV|PronType=Int": 29,
+    "I-ADV|_": 30,
+    "I-AUX|_": 31,
+    "I-CCONJ|_": 32,
+    "I-DET|_": 33,
+    "I-INTJ|_": 34,
+    "I-NOUN|_": 35,
+    "I-NUM|_": 36,
+    "I-PART|Aspect=Perf": 37,
+    "I-PART|Aspect=Prog": 38,
+    "I-PART|Polarity=Neg": 39,
+    "I-PART|PronType=Int": 40,
+    "I-PART|_": 41,
+    "I-PRON|Person=1": 42,
+    "I-PRON|Person=2": 43,
+    "I-PRON|Person=3": 44,
+    "I-PRON|_": 45,
+    "I-PROPN|_": 46,
+    "I-PUNCT|_": 47,
+    "I-VERB|Voice=Pass": 48,
+    "I-VERB|_": 49,
+    "I-X|_": 50
+  },
+  "layer_norm_eps": 1e-12,
+  "max_position_embeddings": 512,
+  "model_type": "roberta",
+  "num_attention_heads": 12,
+  "num_hidden_layers": 12,
+  "pad_token_id": 1,
+  "position_embedding_type": "absolute",
+  "tokenizer_class": "RemBertTokenizerFast",
+  "torch_dtype": "float32",
+  "transformers_version": "4.19.4",
+  "type_vocab_size": 2,
+  "use_cache": true,
+  "vocab_size": 3005
+}

tagger/pytorch_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7f651e118f06d2162ea925c32f7764e027782ecbbf1ea980586eb18673e99ae5
+size 351262257

tagger/sentencepiece.model

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:01ba4719c80b6fe911b091a7c05124b64eeece964e09c058ef8f9805daca546b
+size 1

tagger/special_tokens_map.json

@@ -0,0 +1 @@
+{"bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": {"content": "[MASK]", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true}}

tagger/tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": false, "remove_space": true, "keep_accents": true, "bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": {"content": "[MASK]", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "model_max_length": 512, "tokenizer_class": "RemBertTokenizerFast"}

tokenizer_config.json

@@ -0,0 +1 @@
+{"do_lower_case": false, "remove_space": true, "keep_accents": true, "bos_token": "[CLS]", "eos_token": "[SEP]", "unk_token": "[UNK]", "sep_token": "[SEP]", "pad_token": "[PAD]", "cls_token": "[CLS]", "mask_token": {"content": "[MASK]", "single_word": false, "lstrip": true, "rstrip": false, "normalized": true, "__type": "AddedToken"}, "model_max_length": 512, "tokenizer_class": "RemBertTokenizerFast"}