ArneBinder commited on
Commit
cea3572
·
1 Parent(s): 3ddae0f

adjust for pytorch-ie 0.28.0

Browse files
Files changed (2) hide show
  1. requirements.txt +1 -0
  2. tacred.py +38 -46
requirements.txt ADDED
@@ -0,0 +1 @@
 
 
1
+ pie-datasets>=0.3.0
tacred.py CHANGED
@@ -1,12 +1,16 @@
1
  from dataclasses import dataclass, field
2
- from typing import Any, Callable, Dict, List, Optional, Tuple
3
 
4
  import datasets
5
- import pytorch_ie.data.builder
6
- from pytorch_ie import token_based_document_to_text_based
7
- from pytorch_ie.annotations import BinaryRelation, LabeledSpan, _post_init_single_label
8
- from pytorch_ie.core import Annotation, AnnotationList, Document, annotation_field
9
- from pytorch_ie.documents import TextDocumentWithLabeledEntitiesAndRelations, TokenBasedDocument
 
 
 
 
10
 
11
 
12
  @dataclass(eq=True, frozen=True)
@@ -14,10 +18,7 @@ class TokenRelation(Annotation):
14
  head_idx: int
15
  tail_idx: int
16
  label: str
17
- score: float = 1.0
18
-
19
- def __post_init__(self) -> None:
20
- _post_init_single_label(self)
21
 
22
 
23
  @dataclass(eq=True, frozen=True)
@@ -27,10 +28,7 @@ class TokenAttribute(Annotation):
27
 
28
 
29
  @dataclass
30
- class TacredDocument(Document):
31
- tokens: Tuple[str, ...]
32
- id: Optional[str] = None
33
- metadata: Dict[str, Any] = field(default_factory=dict)
34
  stanford_ner: AnnotationList[TokenAttribute] = annotation_field(target="tokens")
35
  stanford_pos: AnnotationList[TokenAttribute] = annotation_field(target="tokens")
36
  entities: AnnotationList[LabeledSpan] = annotation_field(target="tokens")
@@ -40,14 +38,14 @@ class TacredDocument(Document):
40
 
41
  @dataclass
42
  class SimpleTacredDocument(TokenBasedDocument):
43
- entities: AnnotationList[LabeledSpan] = annotation_field(target="tokens")
44
- relations: AnnotationList[BinaryRelation] = annotation_field(target="entities")
45
 
46
 
47
  def example_to_document(
48
  example: Dict[str, Any],
49
- relation_int2str: Callable[[int], str],
50
- ner_int2str: Callable[[int], str],
51
  ) -> TacredDocument:
52
  document = TacredDocument(
53
  tokens=tuple(example["token"]), id=example["id"], metadata=dict(doc_id=example["docid"])
@@ -72,17 +70,17 @@ def example_to_document(
72
  head = LabeledSpan(
73
  start=example["subj_start"],
74
  end=example["subj_end"],
75
- label=ner_int2str(example["subj_type"]),
76
  )
77
  tail = LabeledSpan(
78
  start=example["obj_start"],
79
  end=example["obj_end"],
80
- label=ner_int2str(example["obj_type"]),
81
  )
82
  document.entities.append(head)
83
  document.entities.append(tail)
84
 
85
- relation_str = relation_int2str(example["relation"])
86
  relation = BinaryRelation(head=head, tail=tail, label=relation_str)
87
  document.relations.append(relation)
88
 
@@ -90,30 +88,20 @@ def example_to_document(
90
 
91
 
92
  def _entity_to_dict(
93
- entity: LabeledSpan, key_prefix: str = "", label_mapping: Optional[Dict[str, Any]] = None
94
  ) -> Dict[str, Any]:
95
  return {
96
  f"{key_prefix}start": entity.start,
97
  f"{key_prefix}end": entity.end,
98
- f"{key_prefix}type": label_mapping[entity.label]
99
- if label_mapping is not None
100
- else entity.label,
101
  }
102
 
103
 
104
  def document_to_example(
105
  document: TacredDocument,
106
- ner_names: Optional[List[str]] = None,
107
- relation_names: Optional[List[str]] = None,
108
  ) -> Dict[str, Any]:
109
-
110
- ner2idx = {name: idx for idx, name in enumerate(ner_names)} if ner_names is not None else None
111
- rel2idx = (
112
- {name: idx for idx, name in enumerate(relation_names)}
113
- if relation_names is not None
114
- else None
115
- )
116
-
117
  token = list(document.tokens)
118
  stanford_ner_dict = {ner.idx: ner.label for ner in document.stanford_ner}
119
  stanford_pos_dict = {pos.idx: pos.label for pos in document.stanford_pos}
@@ -132,24 +120,27 @@ def document_to_example(
132
  return {
133
  "id": document.id,
134
  "docid": document.metadata["doc_id"],
135
- "relation": rel.label if rel2idx is None else rel2idx[rel.label],
136
  "token": token,
137
  "stanford_ner": stanford_ner,
138
  "stanford_pos": stanford_pos,
139
  "stanford_deprel": stanford_deprel,
140
  "stanford_head": stanford_head,
141
- **_entity_to_dict(obj, key_prefix="obj_", label_mapping=ner2idx),
142
- **_entity_to_dict(subj, key_prefix="subj_", label_mapping=ner2idx),
143
  }
144
 
145
 
146
- def convert_to_text_document_with_labeled_entities_and_relations(
147
  document: TacredDocument,
148
- ) -> TextDocumentWithLabeledEntitiesAndRelations:
149
- doc_simplified = document.as_type(SimpleTacredDocument)
 
 
 
150
  result = token_based_document_to_text_based(
151
  doc_simplified,
152
- result_document_type=TextDocumentWithLabeledEntitiesAndRelations,
153
  join_tokens_with=" ",
154
  )
155
  return result
@@ -160,17 +151,18 @@ class TacredConfig(datasets.BuilderConfig):
160
 
161
  def __init__(self, **kwargs):
162
  """BuilderConfig for Tacred.
 
163
  Args:
164
  **kwargs: keyword arguments forwarded to super.
165
  """
166
  super().__init__(**kwargs)
167
 
168
 
169
- class Tacred(pytorch_ie.data.builder.GeneratorBasedBuilder):
170
  DOCUMENT_TYPE = TacredDocument
171
 
172
  DOCUMENT_CONVERTERS = {
173
- TextDocumentWithLabeledEntitiesAndRelations: convert_to_text_document_with_labeled_entities_and_relations,
174
  }
175
 
176
  BASE_DATASET_PATH = "DFKI-SLT/tacred"
@@ -193,8 +185,8 @@ class Tacred(pytorch_ie.data.builder.GeneratorBasedBuilder):
193
 
194
  def _generate_document_kwargs(self, dataset):
195
  return {
196
- "ner_int2str": dataset.features["subj_type"].int2str,
197
- "relation_int2str": dataset.features["relation"].int2str,
198
  }
199
 
200
  def _generate_document(self, example, **kwargs):
 
1
  from dataclasses import dataclass, field
2
+ from typing import Any, Dict, Optional
3
 
4
  import datasets
5
+ from pytorch_ie.annotations import BinaryRelation, LabeledSpan
6
+ from pytorch_ie.core import Annotation, AnnotationList, annotation_field
7
+ from pytorch_ie.documents import (
8
+ TextDocumentWithLabeledSpansAndBinaryRelations,
9
+ TokenBasedDocument,
10
+ )
11
+
12
+ from pie_datasets import GeneratorBasedBuilder
13
+ from pie_datasets.document.conversion import token_based_document_to_text_based
14
 
15
 
16
  @dataclass(eq=True, frozen=True)
 
18
  head_idx: int
19
  tail_idx: int
20
  label: str
21
+ score: float = field(default=1.0, compare=False)
 
 
 
22
 
23
 
24
  @dataclass(eq=True, frozen=True)
 
28
 
29
 
30
  @dataclass
31
+ class TacredDocument(TokenBasedDocument):
 
 
 
32
  stanford_ner: AnnotationList[TokenAttribute] = annotation_field(target="tokens")
33
  stanford_pos: AnnotationList[TokenAttribute] = annotation_field(target="tokens")
34
  entities: AnnotationList[LabeledSpan] = annotation_field(target="tokens")
 
38
 
39
  @dataclass
40
  class SimpleTacredDocument(TokenBasedDocument):
41
+ labeled_spans: AnnotationList[LabeledSpan] = annotation_field(target="tokens")
42
+ binary_relations: AnnotationList[BinaryRelation] = annotation_field(target="labeled_spans")
43
 
44
 
45
  def example_to_document(
46
  example: Dict[str, Any],
47
+ relation_labels: datasets.ClassLabel,
48
+ ner_labels: datasets.ClassLabel,
49
  ) -> TacredDocument:
50
  document = TacredDocument(
51
  tokens=tuple(example["token"]), id=example["id"], metadata=dict(doc_id=example["docid"])
 
70
  head = LabeledSpan(
71
  start=example["subj_start"],
72
  end=example["subj_end"],
73
+ label=ner_labels.int2str(example["subj_type"]),
74
  )
75
  tail = LabeledSpan(
76
  start=example["obj_start"],
77
  end=example["obj_end"],
78
+ label=ner_labels.int2str(example["obj_type"]),
79
  )
80
  document.entities.append(head)
81
  document.entities.append(tail)
82
 
83
+ relation_str = relation_labels.int2str(example["relation"])
84
  relation = BinaryRelation(head=head, tail=tail, label=relation_str)
85
  document.relations.append(relation)
86
 
 
88
 
89
 
90
  def _entity_to_dict(
91
+ entity: LabeledSpan, key_prefix: str = "", labels: Optional[datasets.ClassLabel] = None
92
  ) -> Dict[str, Any]:
93
  return {
94
  f"{key_prefix}start": entity.start,
95
  f"{key_prefix}end": entity.end,
96
+ f"{key_prefix}type": labels.str2int(entity.label) if labels is not None else entity.label,
 
 
97
  }
98
 
99
 
100
  def document_to_example(
101
  document: TacredDocument,
102
+ ner_labels: Optional[datasets.ClassLabel] = None,
103
+ relation_labels: Optional[datasets.ClassLabel] = None,
104
  ) -> Dict[str, Any]:
 
 
 
 
 
 
 
 
105
  token = list(document.tokens)
106
  stanford_ner_dict = {ner.idx: ner.label for ner in document.stanford_ner}
107
  stanford_pos_dict = {pos.idx: pos.label for pos in document.stanford_pos}
 
120
  return {
121
  "id": document.id,
122
  "docid": document.metadata["doc_id"],
123
+ "relation": rel.label if relation_labels is None else relation_labels.str2int(rel.label),
124
  "token": token,
125
  "stanford_ner": stanford_ner,
126
  "stanford_pos": stanford_pos,
127
  "stanford_deprel": stanford_deprel,
128
  "stanford_head": stanford_head,
129
+ **_entity_to_dict(obj, key_prefix="obj_", labels=ner_labels),
130
+ **_entity_to_dict(subj, key_prefix="subj_", labels=ner_labels),
131
  }
132
 
133
 
134
+ def convert_to_text_document_with_labeled_spans_and_binary_relations(
135
  document: TacredDocument,
136
+ ) -> TextDocumentWithLabeledSpansAndBinaryRelations:
137
+ doc_simplified = document.as_type(
138
+ SimpleTacredDocument,
139
+ field_mapping={"entities": "labeled_spans", "relations": "binary_relations"},
140
+ )
141
  result = token_based_document_to_text_based(
142
  doc_simplified,
143
+ result_document_type=TextDocumentWithLabeledSpansAndBinaryRelations,
144
  join_tokens_with=" ",
145
  )
146
  return result
 
151
 
152
  def __init__(self, **kwargs):
153
  """BuilderConfig for Tacred.
154
+
155
  Args:
156
  **kwargs: keyword arguments forwarded to super.
157
  """
158
  super().__init__(**kwargs)
159
 
160
 
161
+ class Tacred(GeneratorBasedBuilder):
162
  DOCUMENT_TYPE = TacredDocument
163
 
164
  DOCUMENT_CONVERTERS = {
165
+ TextDocumentWithLabeledSpansAndBinaryRelations: convert_to_text_document_with_labeled_spans_and_binary_relations,
166
  }
167
 
168
  BASE_DATASET_PATH = "DFKI-SLT/tacred"
 
185
 
186
  def _generate_document_kwargs(self, dataset):
187
  return {
188
+ "ner_labels": dataset.features["subj_type"],
189
+ "relation_labels": dataset.features["relation"],
190
  }
191
 
192
  def _generate_document(self, example, **kwargs):