forked from lancedb/lancedb
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathgte.py
More file actions
131 lines (111 loc) · 4.16 KB
/
gte.py
File metadata and controls
131 lines (111 loc) · 4.16 KB
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
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
# Copyright (c) 2023. LanceDB Developers
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
from typing import List, Union
import numpy as np
from ..util import attempt_import_or_raise
from .base import TextEmbeddingFunction
from .registry import register
from .utils import weak_lru
@register("gte-text")
class GteEmbeddings(TextEmbeddingFunction):
"""
An embedding function that uses GTE-LARGE MLX format(for Apple silicon devices only)
as well as the standard cpu/gpu version from: https://huggingface.co/thenlper/gte-large.
For Apple users, you will need the mlx package insalled, which can be done with:
pip install mlx
Parameters
----------
name: str, default "thenlper/gte-large"
The name of the model to use.
device: str, default "cpu"
Sets the device type for the model.
normalize: str, default "True"
Controls normalize param in encode function for the transformer.
mlx: bool, default False
Controls which model to use. False for gte-large,True for the mlx version.
Examples
--------
import lancedb
import lancedb.embeddings.gte
from lancedb.embeddings import get_registry
from lancedb.pydantic import LanceModel, Vector
import pandas as pd
model = get_registry().get("gte-text").create() # mlx=True for Apple silicon
class TextModel(LanceModel):
text: str = model.SourceField()
vector: Vector(model.ndims()) = model.VectorField()
df = pd.DataFrame({"text": ["hi hello sayonara", "goodbye world"]})
db = lancedb.connect("~/.lancedb")
tbl = db.create_table("test", schema=TextModel, mode="overwrite")
tbl.add(df)
rs = tbl.search("hello").limit(1).to_pandas()
"""
name: str = "thenlper/gte-large"
device: str = "cpu"
normalize: bool = True
mlx: bool = False
def __init__(self, **kwargs):
super().__init__(**kwargs)
self._ndims = None
if kwargs:
self.mlx = kwargs.get("mlx", False)
if self.mlx is True:
self.name = "gte-mlx"
@property
def embedding_model(self):
"""
Get the embedding model specified by the flag,
name and device. This is cached so that the model is only loaded
once per process.
"""
return self.get_embedding_model()
def ndims(self):
if self.mlx is True:
self._ndims = self.embedding_model.dims
if self._ndims is None:
self._ndims = len(self.generate_embeddings("foo")[0])
return self._ndims
def generate_embeddings(
self, texts: Union[List[str], np.ndarray]
) -> List[np.array]:
"""
Get the embeddings for the given texts.
Parameters
----------
texts: list[str] or np.ndarray (of str)
The texts to embed
"""
if self.mlx is True:
return self.embedding_model.run(list(texts)).tolist()
return self.embedding_model.encode(
list(texts),
convert_to_numpy=True,
normalize_embeddings=self.normalize,
).tolist()
@weak_lru(maxsize=1)
def get_embedding_model(self):
"""
Get the embedding model specified by the flag,
name and device. This is cached so that the model is only loaded
once per process.
"""
if self.mlx is True:
from .gte_mlx_model import Model
return Model()
else:
sentence_transformers = attempt_import_or_raise(
"sentence_transformers", "sentence-transformers"
)
return sentence_transformers.SentenceTransformer(
self.name, device=self.device
)