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 | class DeepInfraEmbeddingModel(BaseEmbedding):
"""
A wrapper class for accessing embedding models available via the DeepInfra API. This class allows for easy integration
of DeepInfra embeddings into your projects, supporting both synchronous and asynchronous retrieval of text embeddings.
Args:
model_id (str): Identifier for the model to be used for embeddings. Defaults to 'sentence-transformers/clip-ViT-B-32'.
normalize (bool): Flag to normalize embeddings post retrieval. Defaults to False.
api_token (str): DeepInfra API token. If not provided,
the token is fetched from the environment variable 'DEEPINFRA_API_TOKEN'.
Examples:
>>> from llama_index.embeddings.deepinfra import DeepInfraEmbeddingModel
>>> model = DeepInfraEmbeddingModel()
>>> print(model.get_text_embedding("Hello, world!"))
[0.1, 0.2, 0.3, ...]
"""
"""model_id can be obtained from the DeepInfra website."""
_model_id: str = PrivateAttr()
"""normalize flag to normalize embeddings post retrieval."""
_normalize: bool = PrivateAttr()
"""api_token should be obtained from the DeepInfra website."""
_api_token: str = PrivateAttr()
"""query_prefix is used to add a prefix to queries."""
_query_prefix: str = PrivateAttr()
"""text_prefix is used to add a prefix to texts."""
_text_prefix: str = PrivateAttr()
def __init__(
self,
model_id: str = DEFAULT_MODEL_ID,
normalize: bool = False,
api_token: str = None,
callback_manager: Optional[CallbackManager] = None,
query_prefix: str = "",
text_prefix: str = "",
embed_batch_size: int = MAX_BATCH_SIZE,
) -> None:
"""
Init params.
"""
super().__init__(
callback_manager=callback_manager, embed_batch_size=embed_batch_size
)
self._model_id = model_id
self._normalize = normalize
self._api_token = api_token or os.getenv(ENV_VARIABLE, None)
self._query_prefix = query_prefix
self._text_prefix = text_prefix
def _post(self, data: List[str]) -> List[List[float]]:
"""
Sends a POST request to the DeepInfra Inference API with the given data and returns the API response.
Input data is chunked into batches to avoid exceeding the maximum batch size (1024).
Args:
data (List[str]): A list of strings to be embedded.
Returns:
dict: A dictionary containing embeddings from the API.
"""
url = self.get_url()
chunked_data = _chunk(data, self.embed_batch_size)
embeddings = []
for chunk in chunked_data:
response = requests.post(
url,
json={
"inputs": chunk,
},
headers=self._get_headers(),
)
response.raise_for_status()
embeddings.extend(response.json()["embeddings"])
return embeddings
def get_url(self):
"""
Get DeepInfra API URL.
"""
return f"{INFERENCE_URL}/{self._model_id}"
async def _apost(self, data: List[str]) -> List[List[float]]:
"""
Sends a POST request to the DeepInfra Inference API with the given data and returns the API response.
Input data is chunked into batches to avoid exceeding the maximum batch size (1024).
Args:
data (List[str]): A list of strings to be embedded.
Output:
List[float]: A list of embeddings from the API.
"""
url = self.get_url()
chunked_data = _chunk(data, self.embed_batch_size)
embeddings = []
for chunk in chunked_data:
async with aiohttp.ClientSession() as session:
async with session.post(
url,
json={
"inputs": chunk,
},
headers=self._get_headers(),
) as resp:
response = await resp.json()
embeddings.extend(response["embeddings"])
return embeddings
def _get_query_embedding(self, query: str) -> List[float]:
"""
Get query embedding.
"""
return self._post(self._add_query_prefix([query]))[0]
async def _aget_query_embedding(self, query: str) -> List[float]:
"""
Async get query embedding.
"""
response = await self._apost(self._add_query_prefix([query]))
return response[0]
def _get_query_embeddings(self, queries: List[str]) -> List[List[float]]:
"""
Get query embeddings.
"""
return self._post(self._add_query_prefix(queries))
async def _aget_query_embeddings(self, queries: List[str]) -> List[List[float]]:
"""
Async get query embeddings.
"""
return await self._apost(self._add_query_prefix(queries))
def _get_text_embedding(self, text: str) -> List[float]:
"""
Get text embedding.
"""
return self._post(self._add_text_prefix([text]))[0]
async def _aget_text_embedding(self, text: str) -> List[float]:
"""
Async get text embedding.
"""
response = await self._apost(self._add_text_prefix([text]))
return response[0]
def _get_text_embeddings(self, texts: List[str]) -> List[List[float]]:
"""
Get text embedding.
"""
return self._post(self._add_text_prefix(texts))
async def _aget_text_embeddings(self, texts: List[str]) -> List[List[float]]:
"""
Async get text embeddings.
"""
return await self._apost(self._add_text_prefix(texts))
def _add_query_prefix(self, queries: List[str]) -> List[str]:
"""
Add query prefix to queries.
"""
return (
[self._query_prefix + query for query in queries]
if self._query_prefix
else queries
)
def _add_text_prefix(self, texts: List[str]) -> List[str]:
"""
Add text prefix to texts.
"""
return (
[self._text_prefix + text for text in texts] if self._text_prefix else texts
)
def _get_headers(self) -> dict:
"""
Get headers.
"""
return {
"Authorization": f"Bearer {self._api_token}",
"Content-Type": "application/json",
"User-Agent": USER_AGENT,
}
|
