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
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416 | class BedrockEmbedding(BaseEmbedding):
model: str = Field(description="The modelId of the Bedrock model to use.")
profile_name: Optional[str] = Field(
description="The name of aws profile to use. If not given, then the default profile is used.",
exclude=True,
)
aws_access_key_id: Optional[str] = Field(
description="AWS Access Key ID to use", exclude=True
)
aws_secret_access_key: Optional[str] = Field(
description="AWS Secret Access Key to use", exclude=True
)
aws_session_token: Optional[str] = Field(
description="AWS Session Token to use", exclude=True
)
region_name: Optional[str] = Field(
description="AWS region name to use. Uses region configured in AWS CLI if not passed",
exclude=True,
)
botocore_session: Optional[Any] = Field(
description="Use this Botocore session instead of creating a new default one.",
exclude=True,
)
botocore_config: Optional[Any] = Field(
description="Custom configuration object to use instead of the default generated one.",
exclude=True,
)
max_retries: int = Field(
default=10, description="The maximum number of API retries.", gt=0
)
timeout: float = Field(
default=60.0,
description="The timeout for the Bedrock API request in seconds. It will be used for both connect and read timeouts.",
)
additional_kwargs: Dict[str, Any] = Field(
default_factory=dict, description="Additional kwargs for the bedrock client."
)
_client: Any = PrivateAttr()
def __init__(
self,
model: str = Models.TITAN_EMBEDDING,
profile_name: Optional[str] = None,
aws_access_key_id: Optional[str] = None,
aws_secret_access_key: Optional[str] = None,
aws_session_token: Optional[str] = None,
region_name: Optional[str] = None,
client: Optional[Any] = None,
botocore_session: Optional[Any] = None,
botocore_config: Optional[Any] = None,
additional_kwargs: Optional[Dict[str, Any]] = None,
max_retries: int = 10,
timeout: float = 60.0,
callback_manager: Optional[CallbackManager] = None,
# base class
system_prompt: Optional[str] = None,
messages_to_prompt: Optional[Callable[[Sequence[ChatMessage]], str]] = None,
completion_to_prompt: Optional[Callable[[str], str]] = None,
pydantic_program_mode: PydanticProgramMode = PydanticProgramMode.DEFAULT,
output_parser: Optional[BaseOutputParser] = None,
**kwargs: Any,
):
additional_kwargs = additional_kwargs or {}
session_kwargs = {
"profile_name": profile_name,
"region_name": region_name,
"aws_access_key_id": aws_access_key_id,
"aws_secret_access_key": aws_secret_access_key,
"aws_session_token": aws_session_token,
"botocore_session": botocore_session,
}
config = None
try:
import boto3
from botocore.config import Config
config = (
Config(
retries={"max_attempts": max_retries, "mode": "standard"},
connect_timeout=timeout,
read_timeout=timeout,
)
if botocore_config is None
else botocore_config
)
session = boto3.Session(**session_kwargs)
except ImportError:
raise ImportError(
"boto3 package not found, install with" "'pip install boto3'"
)
# Prior to general availability, custom boto3 wheel files were
# distributed that used the bedrock service to invokeModel.
# This check prevents any services still using those wheel files
# from breaking
if client is not None:
self._client = client
elif "bedrock-runtime" in session.get_available_services():
self._client = session.client("bedrock-runtime", config=config)
else:
self._client = session.client("bedrock", config=config)
super().__init__(
model=model,
max_retries=max_retries,
timeout=timeout,
botocore_config=config,
profile_name=profile_name,
aws_access_key_id=aws_access_key_id,
aws_secret_access_key=aws_secret_access_key,
aws_session_token=aws_session_token,
region_name=region_name,
botocore_session=botocore_session,
additional_kwargs=additional_kwargs,
callback_manager=callback_manager,
system_prompt=system_prompt,
messages_to_prompt=messages_to_prompt,
completion_to_prompt=completion_to_prompt,
pydantic_program_mode=pydantic_program_mode,
output_parser=output_parser,
**kwargs,
)
@staticmethod
def list_supported_models() -> Dict[str, List[str]]:
list_models = {}
for provider in PROVIDERS:
list_models[provider.value] = [m.value for m in Models]
return list_models
@classmethod
def class_name(self) -> str:
return "BedrockEmbedding"
@deprecated(
version="0.9.48",
reason=(
"Use the provided kwargs in the constructor, "
"set_credentials will be removed in future releases."
),
action="once",
)
def set_credentials(
self,
aws_region: Optional[str] = None,
aws_access_key_id: Optional[str] = None,
aws_secret_access_key: Optional[str] = None,
aws_session_token: Optional[str] = None,
aws_profile: Optional[str] = None,
) -> None:
aws_region = aws_region or os.getenv("AWS_REGION")
aws_access_key_id = aws_access_key_id or os.getenv("AWS_ACCESS_KEY_ID")
aws_secret_access_key = aws_secret_access_key or os.getenv(
"AWS_SECRET_ACCESS_KEY"
)
aws_session_token = aws_session_token or os.getenv("AWS_SESSION_TOKEN")
if aws_region is None:
warnings.warn(
"AWS_REGION not found. Set environment variable AWS_REGION or set aws_region"
)
if aws_access_key_id is None:
warnings.warn(
"AWS_ACCESS_KEY_ID not found. Set environment variable AWS_ACCESS_KEY_ID or set aws_access_key_id"
)
assert aws_access_key_id is not None
if aws_secret_access_key is None:
warnings.warn(
"AWS_SECRET_ACCESS_KEY not found. Set environment variable AWS_SECRET_ACCESS_KEY or set aws_secret_access_key"
)
assert aws_secret_access_key is not None
if aws_session_token is None:
warnings.warn(
"AWS_SESSION_TOKEN not found. Set environment variable AWS_SESSION_TOKEN or set aws_session_token"
)
assert aws_session_token is not None
session_kwargs = {
"profile_name": aws_profile,
"region_name": aws_region,
"aws_access_key_id": aws_access_key_id,
"aws_secret_access_key": aws_secret_access_key,
"aws_session_token": aws_session_token,
}
try:
import boto3
session = boto3.Session(**session_kwargs)
except ImportError:
raise ImportError(
"boto3 package not found, install with" "'pip install boto3'"
)
if "bedrock-runtime" in session.get_available_services():
self._client = session.client("bedrock-runtime")
else:
self._client = session.client("bedrock")
@classmethod
@deprecated(
version="0.9.48",
reason=(
"Use the provided kwargs in the constructor, "
"set_credentials will be removed in future releases."
),
action="once",
)
def from_credentials(
cls,
model_name: str = Models.TITAN_EMBEDDING,
aws_region: Optional[str] = None,
aws_access_key_id: Optional[str] = None,
aws_secret_access_key: Optional[str] = None,
aws_session_token: Optional[str] = None,
aws_profile: Optional[str] = None,
embed_batch_size: int = DEFAULT_EMBED_BATCH_SIZE,
callback_manager: Optional[CallbackManager] = None,
verbose: bool = False,
) -> "BedrockEmbedding":
"""
Instantiate using AWS credentials.
Args:
model_name (str) : Name of the model
aws_access_key_id (str): AWS access key ID
aws_secret_access_key (str): AWS secret access key
aws_session_token (str): AWS session token
aws_region (str): AWS region where the service is located
aws_profile (str): AWS profile, when None, default profile is chosen automatically
Example:
.. code-block:: python
from llama_index.embeddings import BedrockEmbedding
# Define the model name
model_name = "your_model_name"
embeddings = BedrockEmbedding.from_credentials(
model_name,
aws_access_key_id,
aws_secret_access_key,
aws_session_token,
aws_region,
aws_profile,
)
"""
session_kwargs = {
"profile_name": aws_profile,
"region_name": aws_region,
"aws_access_key_id": aws_access_key_id,
"aws_secret_access_key": aws_secret_access_key,
"aws_session_token": aws_session_token,
}
try:
import boto3
session = boto3.Session(**session_kwargs)
except ImportError:
raise ImportError(
"boto3 package not found, install with" "'pip install boto3'"
)
if "bedrock-runtime" in session.get_available_services():
client = session.client("bedrock-runtime")
else:
client = session.client("bedrock")
return cls(
client=client,
model=model_name,
embed_batch_size=embed_batch_size,
callback_manager=callback_manager,
verbose=verbose,
)
def _get_embedding(
self, payload: Union[str, List[str]], type: Literal["text", "query"]
) -> Union[Embedding, List[Embedding]]:
"""Get the embedding for the given payload.
Args:
payload (Union[str, List[str]]): The text or list of texts for which the embeddings are to be obtained.
type (Literal["text", "query"]): The type of the payload. It can be either "text" or "query".
Returns:
Union[Embedding, List[Embedding]]: The embedding or list of embeddings for the given payload. If the payload is a list of strings, then the response will be a list of embeddings.
"""
if self._client is None:
self.set_credentials()
if self._client is None:
raise ValueError("Client not set")
provider = self.model.split(".")[0]
request_body = self._get_request_body(provider, payload, type)
response = self._client.invoke_model(
body=request_body,
modelId=self.model,
accept="application/json",
contentType="application/json",
)
resp = json.loads(response.get("body").read().decode("utf-8"))
identifiers = PROVIDER_SPECIFIC_IDENTIFIERS.get(provider, None)
if identifiers is None:
raise ValueError("Provider not supported")
return identifiers["get_embeddings_func"](resp, isinstance(payload, list))
def _get_query_embedding(self, query: str) -> Embedding:
return self._get_embedding(query, "query")
def _get_text_embedding(self, text: str) -> Embedding:
return self._get_embedding(text, "text")
def _get_text_embeddings(self, texts: List[str]) -> List[Embedding]:
provider = self.model.split(".")[0]
if provider == PROVIDERS.COHERE:
return self._get_embedding(texts, "text")
return super()._get_text_embeddings(texts)
def _get_request_body(
self,
provider: str,
payload: Union[str, List[str]],
input_type: Literal["text", "query"],
) -> Any:
"""Build the request body as per the provider.
Currently supported providers are amazon, cohere.
amazon:
Sample Payload of type str
"Hello World!"
cohere:
Sample Payload of type dict of following format
{
'texts': ["This is a test document", "This is another document"],
'input_type': 'search_document',
'truncate': 'NONE'
}
"""
if provider == PROVIDERS.AMAZON:
if isinstance(payload, list):
raise ValueError("Amazon provider does not support list of texts")
request_body = json.dumps({"inputText": payload})
elif provider == PROVIDERS.COHERE:
input_types = {
"text": "search_document",
"query": "search_query",
}
payload = [payload] if isinstance(payload, str) else payload
payload = [p[:2048] if len(p) > 2048 else p for p in payload]
request_body = json.dumps(
{
"texts": payload,
"input_type": input_types[input_type],
"truncate": "NONE",
}
)
else:
raise ValueError("Provider not supported")
return request_body
async def _aget_query_embedding(self, query: str) -> Embedding:
return self._get_embedding(query, "query")
async def _aget_text_embedding(self, text: str) -> Embedding:
return self._get_embedding(text, "text")
|