feat: Adds handler.py for inference endpoints
Browse files- README.md +5 -0
- handler.py +46 -0
- requirements.txt +1 -0
README.md
CHANGED
@@ -15,6 +15,11 @@ widget:
|
|
15 |
- text: "George Washington ging nach Washington"
|
16 |
---
|
17 |
|
|
|
|
|
|
|
|
|
|
|
18 |
## 4-Language NER in Flair (English, German, Dutch and Spanish)
|
19 |
|
20 |
This is the standard 4-class NER model for 4 CoNLL-03 languages that ships with [Flair](https://github.com/flairNLP/flair/). Also kind of works for related languages like French.
|
|
|
15 |
- text: "George Washington ging nach Washington"
|
16 |
---
|
17 |
|
18 |
+
# This is a fork of flair/ner-multi
|
19 |
+
|
20 |
+
As `flair/ner-multi` is missing a `handler.py`, this form implements a custom `handler.py` to be used with inference endpoints. The original model can be found here: [https://huggingface.co/flair/ner-multi](https://huggingface.co/flair/ner-multi)
|
21 |
+
|
22 |
+
|
23 |
## 4-Language NER in Flair (English, German, Dutch and Spanish)
|
24 |
|
25 |
This is the standard 4-class NER model for 4 CoNLL-03 languages that ships with [Flair](https://github.com/flairNLP/flair/). Also kind of works for related languages like French.
|
handler.py
ADDED
@@ -0,0 +1,46 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from typing import Any, Dict, List
|
2 |
+
import os
|
3 |
+
from flair.data import Sentence
|
4 |
+
from flair.models import SequenceTagger
|
5 |
+
|
6 |
+
class EndpointHandler():
|
7 |
+
def __init__(
|
8 |
+
self,
|
9 |
+
path: str,
|
10 |
+
):
|
11 |
+
self.tagger = SequenceTagger.load(os.path.join(path,"pytorch_model.bin"))
|
12 |
+
|
13 |
+
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
14 |
+
"""
|
15 |
+
Args:
|
16 |
+
inputs (:obj:`str`):
|
17 |
+
a string containing some text
|
18 |
+
Return:
|
19 |
+
A :obj:`list`:. The object returned should be like [{"entity_group": "XXX", "word": "some word", "start": 3, "end": 6, "score": 0.82}] containing :
|
20 |
+
- "entity_group": A string representing what the entity is.
|
21 |
+
- "word": A substring of the original string that was detected as an entity.
|
22 |
+
- "start": the offset within `input` leading to `answer`. context[start:stop] == word
|
23 |
+
- "end": the ending offset within `input` leading to `answer`. context[start:stop] === word
|
24 |
+
- "score": A score between 0 and 1 describing how confident the model is for this entity.
|
25 |
+
"""
|
26 |
+
inputs = data.pop("inputs", data)
|
27 |
+
sentence: Sentence = Sentence(inputs)
|
28 |
+
|
29 |
+
# Also show scores for recognized NEs
|
30 |
+
self.tagger.predict(sentence, label_name="predicted")
|
31 |
+
|
32 |
+
entities = []
|
33 |
+
for span in sentence.get_spans("predicted"):
|
34 |
+
if len(span.tokens) == 0:
|
35 |
+
continue
|
36 |
+
current_entity = {
|
37 |
+
"entity_group": span.tag,
|
38 |
+
"word": span.text,
|
39 |
+
"start": span.tokens[0].start_position,
|
40 |
+
"end": span.tokens[-1].end_position,
|
41 |
+
"score": span.score,
|
42 |
+
}
|
43 |
+
|
44 |
+
entities.append(current_entity)
|
45 |
+
|
46 |
+
return entities
|
requirements.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
flair==0.12.2
|