forked from lancedb/lancedb
-
Notifications
You must be signed in to change notification settings - Fork 0
Expand file tree
/
Copy pathbedrock.py
More file actions
235 lines (200 loc) · 8.03 KB
/
bedrock.py
File metadata and controls
235 lines (200 loc) · 8.03 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
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
# 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.
import json
from functools import cached_property
from typing import List, Union
import numpy as np
from lancedb.pydantic import PYDANTIC_VERSION
from ..util import attempt_import_or_raise
from .base import TextEmbeddingFunction
from .registry import register
from .utils import TEXT
@register("bedrock-text")
class BedRockText(TextEmbeddingFunction):
"""
Parameters
----------
name: str, default "amazon.titan-embed-text-v1"
The model ID of the bedrock model to use. Supported models for are:
- amazon.titan-embed-text-v1
- cohere.embed-english-v3
- cohere.embed-multilingual-v3
region: str, default "us-east-1"
Optional name of the AWS Region in which the service should be called.
profile_name: str, default None
Optional name of the AWS profile to use for calling the Bedrock service.
If not specified, the default profile will be used.
assumed_role: str, default None
Optional ARN of an AWS IAM role to assume for calling the Bedrock service.
If not specified, the current active credentials will be used.
role_session_name: str, default "lancedb-embeddings"
Optional name of the AWS IAM role session to use for calling the Bedrock
service. If not specified, "lancedb-embeddings" name will be used.
Examples
--------
import lancedb
import pandas as pd
from lancedb.pydantic import LanceModel, Vector
model = get_registry().get("bedrock-text").create()
class TextModel(LanceModel):
text: str = model.SourceField()
vector: Vector(model.ndims()) = model.VectorField()
df = pd.DataFrame({"text": ["hello world", "goodbye world"]})
db = lancedb.connect("tmp_path")
tbl = db.create_table("test", schema=TextModel, mode="overwrite")
tbl.add(df)
rs = tbl.search("hello").limit(1).to_pandas()
"""
name: str = "amazon.titan-embed-text-v1"
region: str = "us-east-1"
assumed_role: Union[str, None] = None
profile_name: Union[str, None] = None
role_session_name: str = "lancedb-embeddings"
source_input_type: str = "search_document"
query_input_type: str = "search_query"
if PYDANTIC_VERSION.major < 2: # Pydantic 1.x compat
class Config:
keep_untouched = (cached_property,)
else:
model_config = dict()
model_config["ignored_types"] = (cached_property,)
def ndims(self):
# return len(self._generate_embedding("test"))
# TODO: fix hardcoding
if self.name == "amazon.titan-embed-text-v1":
return 1536
elif self.name in [
"amazon.titan-embed-text-v2:0",
"cohere.embed-english-v3",
"cohere.embed-multilingual-v3",
]:
# TODO: "amazon.titan-embed-text-v2:0" model supports dynamic ndims
return 1024
else:
raise ValueError(f"Model {self.name} not supported")
def compute_query_embeddings(
self, query: str, *args, **kwargs
) -> List[List[float]]:
return self.compute_source_embeddings(query, input_type=self.query_input_type)
def compute_source_embeddings(
self, texts: TEXT, *args, **kwargs
) -> List[List[float]]:
texts = self.sanitize_input(texts)
# assume source input type if not passed by `compute_query_embeddings`
kwargs["input_type"] = kwargs.get("input_type") or self.source_input_type
return self.generate_embeddings(texts, **kwargs)
def generate_embeddings(
self, texts: Union[List[str], np.ndarray], *args, **kwargs
) -> List[List[float]]:
"""
Get the embeddings for the given texts
Parameters
----------
texts: list[str] or np.ndarray (of str)
The texts to embed
Returns
-------
list[list[float]]
The embeddings for the given texts
"""
results = []
for text in texts:
response = self._generate_embedding(text, *args, **kwargs)
results.append(response)
return results
def _generate_embedding(self, text: str, *args, **kwargs) -> List[float]:
"""
Get the embeddings for the given texts
Parameters
----------
texts: str
The texts to embed
Returns
-------
list[float]
The embeddings for the given texts
"""
# format input body for provider
provider = self.name.split(".")[0]
input_body = {**kwargs}
if provider == "cohere":
input_body["texts"] = [text]
else:
# includes common provider == "amazon"
input_body.pop("input_type", None)
input_body["inputText"] = text
body = json.dumps(input_body)
try:
# invoke bedrock API
response = self.client.invoke_model(
body=body,
modelId=self.name,
accept="application/json",
contentType="application/json",
)
# format output based on provider
response_body = json.loads(response.get("body").read())
if provider == "cohere":
return response_body.get("embeddings")[0]
else:
# includes common provider == "amazon"
return response_body.get("embedding")
except Exception as e:
help_txt = """
boto3 client failed to invoke the bedrock API. In case of
AWS credentials error:
- Please check your AWS credentials and ensure that you have access.
You can set up aws credentials using `aws configure` command and
verify by running `aws sts get-caller-identity` in your terminal.
"""
raise ValueError(f"Error raised by boto3 client: {e}. \n {help_txt}")
@cached_property
def client(self):
"""Create a boto3 client for Amazon Bedrock service
Returns
-------
boto3.client
The boto3 client for Amazon Bedrock service
"""
botocore = attempt_import_or_raise("botocore")
boto3 = attempt_import_or_raise("boto3")
session_kwargs = {"region_name": self.region}
client_kwargs = {**session_kwargs}
if self.profile_name:
session_kwargs["profile_name"] = self.profile_name
retry_config = botocore.config.Config(
region_name=self.region,
retries={
"max_attempts": 0, # disable this as retries retries are handled
"mode": "standard",
},
)
session = (
boto3.Session(**session_kwargs) if self.profile_name else boto3.Session()
)
if self.assumed_role: # if not using default credentials
sts = session.client("sts")
response = sts.assume_role(
RoleArn=str(self.assumed_role),
RoleSessionName=self.role_session_name,
)
client_kwargs["aws_access_key_id"] = response["Credentials"]["AccessKeyId"]
client_kwargs["aws_secret_access_key"] = response["Credentials"][
"SecretAccessKey"
]
client_kwargs["aws_session_token"] = response["Credentials"]["SessionToken"]
service_name = "bedrock-runtime"
bedrock_client = session.client(
service_name=service_name, config=retry_config, **client_kwargs
)
return bedrock_client