Sentence Similarity
sentence-transformers
PyTorch
Transformers
English
t5
text-embedding
embeddings
information-retrieval
beir
text-classification
language-model
text-clustering
text-semantic-similarity
text-evaluation
prompt-retrieval
text-reranking
feature-extraction
English
Sentence Similarity
natural_questions
ms_marco
fever
hotpot_qa
mteb
Eval Results
Inference Endpoints
add handler
Browse files- handler.py +32 -0
- requirements.txt +1 -0
handler.py
ADDED
@@ -0,0 +1,32 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from typing import Dict, List, Any
|
2 |
+
from InstructorEmbedding import INSTRUCTOR
|
3 |
+
|
4 |
+
|
5 |
+
INSTRUCTION_SEPARATOR = "|||"
|
6 |
+
|
7 |
+
|
8 |
+
class EndpointHandler:
|
9 |
+
def __init__(self, path=""):
|
10 |
+
# load model
|
11 |
+
self.model = INSTRUCTOR(path, device="cuda")
|
12 |
+
|
13 |
+
def __call__(self, data: Dict[str, Any]) -> List[Dict[str, Any]]:
|
14 |
+
"""
|
15 |
+
data args:
|
16 |
+
inputs (:obj: `str`)
|
17 |
+
Return:
|
18 |
+
A :obj:`list` | `dict`: will be serialized and returned
|
19 |
+
"""
|
20 |
+
# get inputs
|
21 |
+
texts = data.pop("texts", data)
|
22 |
+
instruction = data.pop("instruction", "Represent this sentence:")
|
23 |
+
# if isinstance(inputs, str):
|
24 |
+
# inputs = [inputs]
|
25 |
+
|
26 |
+
# run normal prediction
|
27 |
+
# scores = self.model.predict_proba(inputs)[0]
|
28 |
+
|
29 |
+
# return [{"label": self.id2label[i], "score": score.item()} for i, score in enumerate(scores)]
|
30 |
+
instructions = [[instruction, text] for text in texts]
|
31 |
+
embeddings = self.model.encode(instructions)
|
32 |
+
return embeddings.tolist()
|
requirements.txt
ADDED
@@ -0,0 +1 @@
|
|
|
|
|
1 |
+
InstructorEmbedding~=1.0.1
|