Spaces:
Runtime error
Runtime error
File size: 12,458 Bytes
a6c26b1 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 |
"""Langchain Wrapper around Sambanova embedding APIs."""
import json
from typing import Dict, Generator, List, Optional
import requests
from langchain_core.embeddings import Embeddings
from langchain_core.pydantic_v1 import BaseModel
from langchain_core.utils import get_from_dict_or_env, pre_init
class SambaStudioEmbeddings(BaseModel, Embeddings):
"""SambaNova embedding models.
To use, you should have the environment variables
``SAMBASTUDIO_EMBEDDINGS_BASE_URL``, ``SAMBASTUDIO_EMBEDDINGS_BASE_URI``
``SAMBASTUDIO_EMBEDDINGS_PROJECT_ID``, ``SAMBASTUDIO_EMBEDDINGS_ENDPOINT_ID``,
``SAMBASTUDIO_EMBEDDINGS_API_KEY``
set with your personal sambastudio variable or pass it as a named parameter
to the constructor.
Example:
.. code-block:: python
from langchain_community.embeddings import SambaStudioEmbeddings
embeddings = SambaStudioEmbeddings(sambastudio_embeddings_base_url=base_url,
sambastudio_embeddings_base_uri=base_uri,
sambastudio_embeddings_project_id=project_id,
sambastudio_embeddings_endpoint_id=endpoint_id,
sambastudio_embeddings_api_key=api_key,
batch_size=32)
(or)
embeddings = SambaStudioEmbeddings(batch_size=32)
(or)
# CoE example
embeddings = SambaStudioEmbeddings(
batch_size=1,
model_kwargs={
'select_expert':'e5-mistral-7b-instruct'
}
)
"""
sambastudio_embeddings_base_url: str = ''
"""Base url to use"""
sambastudio_embeddings_base_uri: str = ''
"""endpoint base uri"""
sambastudio_embeddings_project_id: str = ''
"""Project id on sambastudio for model"""
sambastudio_embeddings_endpoint_id: str = ''
"""endpoint id on sambastudio for model"""
sambastudio_embeddings_api_key: str = ''
"""sambastudio api key"""
model_kwargs: dict = {}
"""Key word arguments to pass to the model."""
batch_size: int = 32
"""Batch size for the embedding models"""
@pre_init
def validate_environment(cls, values: Dict) -> Dict:
"""Validate that api key and python package exists in environment."""
values['sambastudio_embeddings_base_url'] = get_from_dict_or_env(
values, 'sambastudio_embeddings_base_url', 'SAMBASTUDIO_EMBEDDINGS_BASE_URL'
)
values['sambastudio_embeddings_base_uri'] = get_from_dict_or_env(
values,
'sambastudio_embeddings_base_uri',
'SAMBASTUDIO_EMBEDDINGS_BASE_URI',
default='api/predict/generic',
)
values['sambastudio_embeddings_project_id'] = get_from_dict_or_env(
values,
'sambastudio_embeddings_project_id',
'SAMBASTUDIO_EMBEDDINGS_PROJECT_ID',
)
values['sambastudio_embeddings_endpoint_id'] = get_from_dict_or_env(
values,
'sambastudio_embeddings_endpoint_id',
'SAMBASTUDIO_EMBEDDINGS_ENDPOINT_ID',
)
values['sambastudio_embeddings_api_key'] = get_from_dict_or_env(
values, 'sambastudio_embeddings_api_key', 'SAMBASTUDIO_EMBEDDINGS_API_KEY'
)
return values
def _get_tuning_params(self) -> str:
"""
Get the tuning parameters to use when calling the model
Returns:
The tuning parameters as a JSON string.
"""
if 'api/v2/predict/generic' in self.sambastudio_embeddings_base_uri:
tuning_params_dict = self.model_kwargs
else:
tuning_params_dict = {
k: {'type': type(v).__name__, 'value': str(v)} for k, v in (self.model_kwargs.items())
}
tuning_params = json.dumps(tuning_params_dict)
return tuning_params
def _get_full_url(self, path: str) -> str:
"""
Return the full API URL for a given path.
:param str path: the sub-path
:returns: the full API URL for the sub-path
:rtype: str
"""
return f'{self.sambastudio_embeddings_base_url}/{self.sambastudio_embeddings_base_uri}/{path}' # noqa: E501
def _iterate_over_batches(self, texts: List[str], batch_size: int) -> Generator:
"""Generator for creating batches in the embed documents method
Args:
texts (List[str]): list of strings to embed
batch_size (int, optional): batch size to be used for the embedding model.
Will depend on the RDU endpoint used.
Yields:
List[str]: list (batch) of strings of size batch size
"""
for i in range(0, len(texts), batch_size):
yield texts[i : i + batch_size]
def embed_documents(self, texts: List[str], batch_size: Optional[int] = None) -> List[List[float]]:
"""Returns a list of embeddings for the given sentences.
Args:
texts (`List[str]`): List of texts to encode
batch_size (`int`): Batch size for the encoding
Returns:
`List[np.ndarray]` or `List[tensor]`: List of embeddings
for the given sentences
"""
if batch_size is None:
batch_size = self.batch_size
http_session = requests.Session()
url = self._get_full_url(f'{self.sambastudio_embeddings_project_id}/{self.sambastudio_embeddings_endpoint_id}')
params = json.loads(self._get_tuning_params())
embeddings = []
if 'api/predict/nlp' in self.sambastudio_embeddings_base_uri:
for batch in self._iterate_over_batches(texts, batch_size):
data = {'inputs': batch, 'params': params}
response = http_session.post(
url,
headers={'key': self.sambastudio_embeddings_api_key},
json=data,
)
if response.status_code != 200:
raise RuntimeError(
f'Sambanova /complete call failed with status code '
f'{response.status_code}.\n Details: {response.text}'
)
try:
embedding = response.json()['data']
embeddings.extend(embedding)
except KeyError:
raise KeyError(
"'data' not found in endpoint response",
response.json(),
)
elif 'api/v2/predict/generic' in self.sambastudio_embeddings_base_uri:
for batch in self._iterate_over_batches(texts, batch_size):
items = [{'id': f'item{i}', 'value': item} for i, item in enumerate(batch)]
data = {'items': items, 'params': params}
response = http_session.post(
url,
headers={'key': self.sambastudio_embeddings_api_key},
json=data,
)
if response.status_code != 200:
raise RuntimeError(
f'Sambanova /complete call failed with status code '
f'{response.status_code}.\n Details: {response.text}'
)
try:
embedding = [item['value'] for item in response.json()['items']]
embeddings.extend(embedding)
except KeyError:
raise KeyError(
"'items' not found in endpoint response",
response.json(),
)
elif 'api/predict/generic' in self.sambastudio_embeddings_base_uri:
for batch in self._iterate_over_batches(texts, batch_size):
data = {'instances': batch, 'params': params}
response = http_session.post(
url,
headers={'key': self.sambastudio_embeddings_api_key},
json=data,
)
if response.status_code != 200:
raise RuntimeError(
f'Sambanova /complete call failed with status code '
f'{response.status_code}.\n Details: {response.text}'
)
try:
if params.get('select_expert'):
embedding = response.json()['predictions']
else:
embedding = response.json()['predictions']
embeddings.extend(embedding)
except KeyError:
raise KeyError(
"'predictions' not found in endpoint response",
response.json(),
)
else:
raise ValueError(
f'handling of endpoint uri: {self.sambastudio_embeddings_base_uri} not implemented' # noqa: E501
)
return embeddings
def embed_query(self, text: str) -> List[float]:
"""Returns a list of embeddings for the given sentences.
Args:
sentences (`List[str]`): List of sentences to encode
Returns:
`List[np.ndarray]` or `List[tensor]`: List of embeddings
for the given sentences
"""
http_session = requests.Session()
url = self._get_full_url(f'{self.sambastudio_embeddings_project_id}/{self.sambastudio_embeddings_endpoint_id}')
params = json.loads(self._get_tuning_params())
if 'api/predict/nlp' in self.sambastudio_embeddings_base_uri:
data = {'inputs': [text], 'params': params}
response = http_session.post(
url,
headers={'key': self.sambastudio_embeddings_api_key},
json=data,
)
if response.status_code != 200:
raise RuntimeError(
f'Sambanova /complete call failed with status code '
f'{response.status_code}.\n Details: {response.text}'
)
try:
embedding = response.json()['data'][0]
except KeyError:
raise KeyError(
"'data' not found in endpoint response",
response.json(),
)
elif 'api/v2/predict/generic' in self.sambastudio_embeddings_base_uri:
data = {'items': [{'id': 'item0', 'value': text}], 'params': params}
response = http_session.post(
url,
headers={'key': self.sambastudio_embeddings_api_key},
json=data,
)
if response.status_code != 200:
raise RuntimeError(
f'Sambanova /complete call failed with status code '
f'{response.status_code}.\n Details: {response.text}'
)
try:
embedding = response.json()['items'][0]['value']
except KeyError:
raise KeyError(
"'items' not found in endpoint response",
response.json(),
)
elif 'api/predict/generic' in self.sambastudio_embeddings_base_uri:
data = {'instances': [text], 'params': params}
response = http_session.post(
url,
headers={'key': self.sambastudio_embeddings_api_key},
json=data,
)
if response.status_code != 200:
raise RuntimeError(
f'Sambanova /complete call failed with status code '
f'{response.status_code}.\n Details: {response.text}'
)
try:
if params.get('select_expert'):
embedding = response.json()['predictions'][0]
else:
embedding = response.json()['predictions'][0]
except KeyError:
raise KeyError(
"'predictions' not found in endpoint response",
response.json(),
)
else:
raise ValueError(
f'handling of endpoint uri: {self.sambastudio_embeddings_base_uri} not implemented' # noqa: E501
)
return embedding
|