Commit
·
c14b84e
1
Parent(s):
821fb54
Create handler.py
Browse files- handler.py +36 -0
handler.py
ADDED
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from transformers import AutoTokenizer, AutoModelForSequenceClassification
|
2 |
+
import torch
|
3 |
+
|
4 |
+
|
5 |
+
class EndpointHandler:
|
6 |
+
def __init__(self, path=""):
|
7 |
+
self.tokenizer = AutoTokenizer.from_pretrained(path)
|
8 |
+
self.model = AutoModelForSequenceClassification.from_pretrained(path, num_labels=3)
|
9 |
+
def tokenize(text, topic):
|
10 |
+
return self.tokenizer(
|
11 |
+
topic,
|
12 |
+
text,
|
13 |
+
max_length=384, #512
|
14 |
+
truncation="only_second",
|
15 |
+
return_offsets_mapping=False,
|
16 |
+
padding="max_length",
|
17 |
+
return_tensors='pt'
|
18 |
+
)
|
19 |
+
self.tokenize = tokenize
|
20 |
+
|
21 |
+
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
22 |
+
"""
|
23 |
+
data args:
|
24 |
+
topic (:obj: `str`)
|
25 |
+
text (:obj: `str`)
|
26 |
+
Return:
|
27 |
+
A :obj:`list` | `dict`: will be serialized and returned
|
28 |
+
"""
|
29 |
+
topic = data.pop("topic", data)
|
30 |
+
text = data.pop("text", data)
|
31 |
+
tokenized_inputs = self.tokenize(text, topic)
|
32 |
+
|
33 |
+
output = self.model(**tokenized_inputs)
|
34 |
+
prediction = torch.argmax(output.logits, dim=-1).item()
|
35 |
+
label = self.model.config.id2label[prediction]
|
36 |
+
return label
|