File size: 14,937 Bytes
e3278e4 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 |
"""
Qdrant Semantic Cache implementation
Has 4 methods:
- set_cache
- get_cache
- async_set_cache
- async_get_cache
"""
import ast
import asyncio
import json
from typing import Any
import litellm
from litellm._logging import print_verbose
from .base_cache import BaseCache
class QdrantSemanticCache(BaseCache):
def __init__( # noqa: PLR0915
self,
qdrant_api_base=None,
qdrant_api_key=None,
collection_name=None,
similarity_threshold=None,
quantization_config=None,
embedding_model="text-embedding-ada-002",
host_type=None,
):
import os
from litellm.llms.custom_httpx.http_handler import (
_get_httpx_client,
get_async_httpx_client,
httpxSpecialProvider,
)
from litellm.secret_managers.main import get_secret_str
if collection_name is None:
raise Exception("collection_name must be provided, passed None")
self.collection_name = collection_name
print_verbose(
f"qdrant semantic-cache initializing COLLECTION - {self.collection_name}"
)
if similarity_threshold is None:
raise Exception("similarity_threshold must be provided, passed None")
self.similarity_threshold = similarity_threshold
self.embedding_model = embedding_model
headers = {}
# check if defined as os.environ/ variable
if qdrant_api_base:
if isinstance(qdrant_api_base, str) and qdrant_api_base.startswith(
"os.environ/"
):
qdrant_api_base = get_secret_str(qdrant_api_base)
if qdrant_api_key:
if isinstance(qdrant_api_key, str) and qdrant_api_key.startswith(
"os.environ/"
):
qdrant_api_key = get_secret_str(qdrant_api_key)
qdrant_api_base = (
qdrant_api_base or os.getenv("QDRANT_URL") or os.getenv("QDRANT_API_BASE")
)
qdrant_api_key = qdrant_api_key or os.getenv("QDRANT_API_KEY")
headers = {"Content-Type": "application/json"}
if qdrant_api_key:
headers["api-key"] = qdrant_api_key
if qdrant_api_base is None:
raise ValueError("Qdrant url must be provided")
self.qdrant_api_base = qdrant_api_base
self.qdrant_api_key = qdrant_api_key
print_verbose(f"qdrant semantic-cache qdrant_api_base: {self.qdrant_api_base}")
self.headers = headers
self.sync_client = _get_httpx_client()
self.async_client = get_async_httpx_client(
llm_provider=httpxSpecialProvider.Caching
)
if quantization_config is None:
print_verbose(
"Quantization config is not provided. Default binary quantization will be used."
)
collection_exists = self.sync_client.get(
url=f"{self.qdrant_api_base}/collections/{self.collection_name}/exists",
headers=self.headers,
)
if collection_exists.status_code != 200:
raise ValueError(
f"Error from qdrant checking if /collections exist {collection_exists.text}"
)
if collection_exists.json()["result"]["exists"]:
collection_details = self.sync_client.get(
url=f"{self.qdrant_api_base}/collections/{self.collection_name}",
headers=self.headers,
)
self.collection_info = collection_details.json()
print_verbose(
f"Collection already exists.\nCollection details:{self.collection_info}"
)
else:
if quantization_config is None or quantization_config == "binary":
quantization_params = {
"binary": {
"always_ram": False,
}
}
elif quantization_config == "scalar":
quantization_params = {
"scalar": {"type": "int8", "quantile": 0.99, "always_ram": False}
}
elif quantization_config == "product":
quantization_params = {
"product": {"compression": "x16", "always_ram": False}
}
else:
raise Exception(
"Quantization config must be one of 'scalar', 'binary' or 'product'"
)
new_collection_status = self.sync_client.put(
url=f"{self.qdrant_api_base}/collections/{self.collection_name}",
json={
"vectors": {"size": 1536, "distance": "Cosine"},
"quantization_config": quantization_params,
},
headers=self.headers,
)
if new_collection_status.json()["result"]:
collection_details = self.sync_client.get(
url=f"{self.qdrant_api_base}/collections/{self.collection_name}",
headers=self.headers,
)
self.collection_info = collection_details.json()
print_verbose(
f"New collection created.\nCollection details:{self.collection_info}"
)
else:
raise Exception("Error while creating new collection")
def _get_cache_logic(self, cached_response: Any):
if cached_response is None:
return cached_response
try:
cached_response = json.loads(
cached_response
) # Convert string to dictionary
except Exception:
cached_response = ast.literal_eval(cached_response)
return cached_response
def set_cache(self, key, value, **kwargs):
print_verbose(f"qdrant semantic-cache set_cache, kwargs: {kwargs}")
import uuid
# get the prompt
messages = kwargs["messages"]
prompt = ""
for message in messages:
prompt += message["content"]
# create an embedding for prompt
embedding_response = litellm.embedding(
model=self.embedding_model,
input=prompt,
cache={"no-store": True, "no-cache": True},
)
# get the embedding
embedding = embedding_response["data"][0]["embedding"]
value = str(value)
assert isinstance(value, str)
data = {
"points": [
{
"id": str(uuid.uuid4()),
"vector": embedding,
"payload": {
"text": prompt,
"response": value,
},
},
]
}
self.sync_client.put(
url=f"{self.qdrant_api_base}/collections/{self.collection_name}/points",
headers=self.headers,
json=data,
)
return
def get_cache(self, key, **kwargs):
print_verbose(f"sync qdrant semantic-cache get_cache, kwargs: {kwargs}")
# get the messages
messages = kwargs["messages"]
prompt = ""
for message in messages:
prompt += message["content"]
# convert to embedding
embedding_response = litellm.embedding(
model=self.embedding_model,
input=prompt,
cache={"no-store": True, "no-cache": True},
)
# get the embedding
embedding = embedding_response["data"][0]["embedding"]
data = {
"vector": embedding,
"params": {
"quantization": {
"ignore": False,
"rescore": True,
"oversampling": 3.0,
}
},
"limit": 1,
"with_payload": True,
}
search_response = self.sync_client.post(
url=f"{self.qdrant_api_base}/collections/{self.collection_name}/points/search",
headers=self.headers,
json=data,
)
results = search_response.json()["result"]
if results is None:
return None
if isinstance(results, list):
if len(results) == 0:
return None
similarity = results[0]["score"]
cached_prompt = results[0]["payload"]["text"]
# check similarity, if more than self.similarity_threshold, return results
print_verbose(
f"semantic cache: similarity threshold: {self.similarity_threshold}, similarity: {similarity}, prompt: {prompt}, closest_cached_prompt: {cached_prompt}"
)
if similarity >= self.similarity_threshold:
# cache hit !
cached_value = results[0]["payload"]["response"]
print_verbose(
f"got a cache hit, similarity: {similarity}, Current prompt: {prompt}, cached_prompt: {cached_prompt}"
)
return self._get_cache_logic(cached_response=cached_value)
else:
# cache miss !
return None
pass
async def async_set_cache(self, key, value, **kwargs):
import uuid
from litellm.proxy.proxy_server import llm_model_list, llm_router
print_verbose(f"async qdrant semantic-cache set_cache, kwargs: {kwargs}")
# get the prompt
messages = kwargs["messages"]
prompt = ""
for message in messages:
prompt += message["content"]
# create an embedding for prompt
router_model_names = (
[m["model_name"] for m in llm_model_list]
if llm_model_list is not None
else []
)
if llm_router is not None and self.embedding_model in router_model_names:
user_api_key = kwargs.get("metadata", {}).get("user_api_key", "")
embedding_response = await llm_router.aembedding(
model=self.embedding_model,
input=prompt,
cache={"no-store": True, "no-cache": True},
metadata={
"user_api_key": user_api_key,
"semantic-cache-embedding": True,
"trace_id": kwargs.get("metadata", {}).get("trace_id", None),
},
)
else:
# convert to embedding
embedding_response = await litellm.aembedding(
model=self.embedding_model,
input=prompt,
cache={"no-store": True, "no-cache": True},
)
# get the embedding
embedding = embedding_response["data"][0]["embedding"]
value = str(value)
assert isinstance(value, str)
data = {
"points": [
{
"id": str(uuid.uuid4()),
"vector": embedding,
"payload": {
"text": prompt,
"response": value,
},
},
]
}
await self.async_client.put(
url=f"{self.qdrant_api_base}/collections/{self.collection_name}/points",
headers=self.headers,
json=data,
)
return
async def async_get_cache(self, key, **kwargs):
print_verbose(f"async qdrant semantic-cache get_cache, kwargs: {kwargs}")
from litellm.proxy.proxy_server import llm_model_list, llm_router
# get the messages
messages = kwargs["messages"]
prompt = ""
for message in messages:
prompt += message["content"]
router_model_names = (
[m["model_name"] for m in llm_model_list]
if llm_model_list is not None
else []
)
if llm_router is not None and self.embedding_model in router_model_names:
user_api_key = kwargs.get("metadata", {}).get("user_api_key", "")
embedding_response = await llm_router.aembedding(
model=self.embedding_model,
input=prompt,
cache={"no-store": True, "no-cache": True},
metadata={
"user_api_key": user_api_key,
"semantic-cache-embedding": True,
"trace_id": kwargs.get("metadata", {}).get("trace_id", None),
},
)
else:
# convert to embedding
embedding_response = await litellm.aembedding(
model=self.embedding_model,
input=prompt,
cache={"no-store": True, "no-cache": True},
)
# get the embedding
embedding = embedding_response["data"][0]["embedding"]
data = {
"vector": embedding,
"params": {
"quantization": {
"ignore": False,
"rescore": True,
"oversampling": 3.0,
}
},
"limit": 1,
"with_payload": True,
}
search_response = await self.async_client.post(
url=f"{self.qdrant_api_base}/collections/{self.collection_name}/points/search",
headers=self.headers,
json=data,
)
results = search_response.json()["result"]
if results is None:
kwargs.setdefault("metadata", {})["semantic-similarity"] = 0.0
return None
if isinstance(results, list):
if len(results) == 0:
kwargs.setdefault("metadata", {})["semantic-similarity"] = 0.0
return None
similarity = results[0]["score"]
cached_prompt = results[0]["payload"]["text"]
# check similarity, if more than self.similarity_threshold, return results
print_verbose(
f"semantic cache: similarity threshold: {self.similarity_threshold}, similarity: {similarity}, prompt: {prompt}, closest_cached_prompt: {cached_prompt}"
)
# update kwargs["metadata"] with similarity, don't rewrite the original metadata
kwargs.setdefault("metadata", {})["semantic-similarity"] = similarity
if similarity >= self.similarity_threshold:
# cache hit !
cached_value = results[0]["payload"]["response"]
print_verbose(
f"got a cache hit, similarity: {similarity}, Current prompt: {prompt}, cached_prompt: {cached_prompt}"
)
return self._get_cache_logic(cached_response=cached_value)
else:
# cache miss !
return None
pass
async def _collection_info(self):
return self.collection_info
async def async_set_cache_pipeline(self, cache_list, **kwargs):
tasks = []
for val in cache_list:
tasks.append(self.async_set_cache(val[0], val[1], **kwargs))
await asyncio.gather(*tasks)
|