Spaces:
Runtime error
Runtime error
Commit
·
9d06087
1
Parent(s):
5a9013c
Update app.py
Browse files
app.py
CHANGED
@@ -6,13 +6,13 @@ from prettytable import PrettyTable
|
|
6 |
from pytorch_ie.annotations import LabeledSpan, BinaryRelation
|
7 |
from pytorch_ie.auto import AutoPipeline
|
8 |
from pytorch_ie.core import AnnotationList, annotation_field
|
9 |
-
from pytorch_ie.documents import
|
10 |
|
11 |
from typing import List
|
12 |
|
13 |
|
14 |
@dataclass
|
15 |
-
class ExampleDocument(
|
16 |
entities: AnnotationList[LabeledSpan] = annotation_field(target="text")
|
17 |
relations: AnnotationList[BinaryRelation] = annotation_field(target="entities")
|
18 |
|
@@ -21,20 +21,22 @@ ner_model_name_or_path = "pie/example-ner-spanclf-conll03"
|
|
21 |
re_model_name_or_path = "pie/example-re-textclf-tacred"
|
22 |
|
23 |
ner_pipeline = AutoPipeline.from_pretrained(ner_model_name_or_path, device=-1, num_workers=0)
|
24 |
-
re_pipeline = AutoPipeline.from_pretrained(re_model_name_or_path, device=-1, num_workers=0)
|
25 |
|
26 |
|
27 |
def predict(text):
|
28 |
document = ExampleDocument(text)
|
29 |
|
|
|
30 |
ner_pipeline(document)
|
31 |
|
32 |
-
|
33 |
-
|
34 |
-
|
35 |
-
print(f"
|
36 |
-
document.entities.append(entity)
|
37 |
|
|
|
38 |
re_pipeline(document)
|
39 |
|
40 |
t = PrettyTable()
|
|
|
6 |
from pytorch_ie.annotations import LabeledSpan, BinaryRelation
|
7 |
from pytorch_ie.auto import AutoPipeline
|
8 |
from pytorch_ie.core import AnnotationList, annotation_field
|
9 |
+
from pytorch_ie.documents import TextBasedDocument
|
10 |
|
11 |
from typing import List
|
12 |
|
13 |
|
14 |
@dataclass
|
15 |
+
class ExampleDocument(TextBasedDocument):
|
16 |
entities: AnnotationList[LabeledSpan] = annotation_field(target="text")
|
17 |
relations: AnnotationList[BinaryRelation] = annotation_field(target="entities")
|
18 |
|
|
|
21 |
re_model_name_or_path = "pie/example-re-textclf-tacred"
|
22 |
|
23 |
ner_pipeline = AutoPipeline.from_pretrained(ner_model_name_or_path, device=-1, num_workers=0)
|
24 |
+
re_pipeline = AutoPipeline.from_pretrained(re_model_name_or_path, device=-1, num_workers=0, taskmodule_kwargs=dict(create_relation_candidates=True))
|
25 |
|
26 |
|
27 |
def predict(text):
|
28 |
document = ExampleDocument(text)
|
29 |
|
30 |
+
# execute NER pipeline
|
31 |
ner_pipeline(document)
|
32 |
|
33 |
+
# show predicted entities and promote them from predictions to ground-truth annotations
|
34 |
+
print(f"detected entities:\n")
|
35 |
+
for entity in document.entities.predictions:
|
36 |
+
print(f"{entity}")
|
37 |
+
document.entities.append(entity.copy())
|
38 |
|
39 |
+
# execute RE pipeline
|
40 |
re_pipeline(document)
|
41 |
|
42 |
t = PrettyTable()
|