be52af70
 tangwang
first commit
|
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
|
"""
Text embedding encoder using BGE-M3 model.
Generates 1024-dimensional vectors for text using the BGE-M3 multilingual model.
"""
import sys
import torch
from sentence_transformers import SentenceTransformer
import time
import threading
from modelscope import snapshot_download
from transformers import AutoModel
import os
import numpy as np
from typing import List, Union
class BgeEncoder:
"""
Singleton text encoder using BGE-M3 model.
Thread-safe singleton pattern ensures only one model instance exists.
"""
_instance = None
_lock = threading.Lock()
def __new__(cls, model_dir='Xorbits/bge-m3'):
with cls._lock:
if cls._instance is None:
cls._instance = super(BgeEncoder, cls).__new__(cls)
print(f"[BgeEncoder] Creating a new instance with model directory: {model_dir}")
cls._instance.model = SentenceTransformer(snapshot_download(model_dir))
print("[BgeEncoder] New instance has been created")
return cls._instance
def encode(
self,
sentences: Union[str, List[str]],
normalize_embeddings: bool = True,
device: str = 'cuda',
batch_size: int = 32
) -> np.ndarray:
"""
Encode text into embeddings.
Args:
sentences: Single string or list of strings to encode
normalize_embeddings: Whether to normalize embeddings
device: Device to use ('cuda' or 'cpu')
batch_size: Batch size for encoding
Returns:
numpy array of shape (n, 1024) containing embeddings
"""
# Move model to specified device
if device == 'gpu':
device = 'cuda'
self.model = self.model.to(device)
embeddings = self.model.encode(
sentences,
normalize_embeddings=normalize_embeddings,
device=device,
show_progress_bar=False,
batch_size=batch_size
)
return embeddings
def encode_batch(
self,
texts: List[str],
batch_size: int = 32,
device: str = 'cuda'
) -> np.ndarray:
"""
Encode a batch of texts efficiently.
Args:
texts: List of texts to encode
batch_size: Batch size for processing
device: Device to use
Returns:
numpy array of embeddings
"""
return self.encode(texts, batch_size=batch_size, device=device)
|