github/langchain
1
0
Fork 0
mirror of https://github.com/hwchase17/langchain.git synced 2025-07-05 20:58:25 +00:00
Code Issues Packages Projects Releases Wiki Activity

Improve handling of empty queries for timescale vector (#12393)

Browse Source 
**Description:** Improve handling of empty queries in timescale-vector.
For timescale-vector it is more efficient to get a None embedding when
the embedding has no semantic meaning. It allows timescale-vector to
perform more optimizations. Thus, when the query is empty, use a None
embedding.

 Also pass down constructor arguments to the timescale vector client.

---------

Co-authored-by: Bagatur <baskaryan@gmail.com>
This commit is contained in:
Matvey Arye 2023-10-27 16:55:16 -04:00 committed by GitHub
parent 38cee5fae0
commit 11505f95d3
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
1 changed files with 19 additions and 8 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

27
libs/langchain/langchain/vectorstores/timescalevector.py
View File

@ -79,6 +79,7 @@ class TimescaleVector(VectorStore):
logger: Optional[logging.Logger] = None, logger: Optional[logging.Logger] = None,
relevance_score_fn: Optional[Callable[[float], float]] = None, relevance_score_fn: Optional[Callable[[float], float]] = None,
time_partition_interval: Optional[timedelta] = None, time_partition_interval: Optional[timedelta] = None,
**kwargs: Any,
) -> None: ) -> None:
try: try:
from timescale_vector import client from timescale_vector import client
@ -103,6 +104,7 @@ class TimescaleVector(VectorStore):
self.num_dimensions, self.num_dimensions,
self._distance_strategy.value.lower(), self._distance_strategy.value.lower(),
time_partition_interval=self._time_partition_interval, time_partition_interval=self._time_partition_interval,
**kwargs,
) )
self.async_client = client.Async( self.async_client = client.Async(
self.service_url, self.service_url,
@ -110,6 +112,7 @@ class TimescaleVector(VectorStore):
self.num_dimensions, self.num_dimensions,
self._distance_strategy.value.lower(), self._distance_strategy.value.lower(),
time_partition_interval=self._time_partition_interval, time_partition_interval=self._time_partition_interval,
**kwargs,
) )
self.__post_init__() self.__post_init__()
@ -310,6 +313,13 @@ class TimescaleVector(VectorStore):
texts=texts, embeddings=embeddings, metadatas=metadatas, ids=ids, **kwargs texts=texts, embeddings=embeddings, metadatas=metadatas, ids=ids, **kwargs
) )
def _embed_query(self, query: str) -> Optional[List[float]]:
# an empty query should not be embedded
if query is None or query == "" or query.isspace():
return None
else:
return self.embedding.embed_query(query)
def similarity_search( def similarity_search(
self, self,
query: str, query: str,
@ -328,7 +338,7 @@ class TimescaleVector(VectorStore):
Returns: Returns:
List of Documents most similar to the query. List of Documents most similar to the query.
""" """
embedding = self.embedding.embed_query(text=query) embedding = self._embed_query(query)
return self.similarity_search_by_vector( return self.similarity_search_by_vector(
embedding=embedding, embedding=embedding,
k=k, k=k,
@ -355,7 +365,7 @@ class TimescaleVector(VectorStore):
Returns: Returns:
List of Documents most similar to the query. List of Documents most similar to the query.
""" """
embedding = self.embedding.embed_query(text=query) embedding = self._embed_query(query)
return await self.asimilarity_search_by_vector( return await self.asimilarity_search_by_vector(
embedding=embedding, embedding=embedding,
k=k, k=k,
@ -382,7 +392,7 @@ class TimescaleVector(VectorStore):
Returns: Returns:
List of Documents most similar to the query and score for each List of Documents most similar to the query and score for each
""" """
embedding = self.embedding.embed_query(query) embedding = self._embed_query(query)
docs = self.similarity_search_with_score_by_vector( docs = self.similarity_search_with_score_by_vector(
embedding=embedding, embedding=embedding,
k=k, k=k,
@ -410,7 +420,8 @@ class TimescaleVector(VectorStore):
Returns: Returns:
List of Documents most similar to the query and score for each List of Documents most similar to the query and score for each
""" """
embedding = self.embedding.embed_query(query)
embedding = self._embed_query(query)
return await self.asimilarity_search_with_score_by_vector( return await self.asimilarity_search_with_score_by_vector(
embedding=embedding, embedding=embedding,
k=k, k=k,
@ -445,7 +456,7 @@ class TimescaleVector(VectorStore):
def similarity_search_with_score_by_vector( def similarity_search_with_score_by_vector(
self, self,
embedding: List[float], embedding: Optional[List[float]],
k: int = 4, k: int = 4,
filter: Optional[Union[dict, list]] = None, filter: Optional[Union[dict, list]] = None,
predicates: Optional[Predicates] = None, predicates: Optional[Predicates] = None,
@ -481,7 +492,7 @@ class TimescaleVector(VectorStore):
async def asimilarity_search_with_score_by_vector( async def asimilarity_search_with_score_by_vector(
self, self,
embedding: List[float], embedding: Optional[List[float]],
k: int = 4, k: int = 4,
filter: Optional[Union[dict, list]] = None, filter: Optional[Union[dict, list]] = None,
predicates: Optional[Predicates] = None, predicates: Optional[Predicates] = None,
@ -517,7 +528,7 @@ class TimescaleVector(VectorStore):
def similarity_search_by_vector( def similarity_search_by_vector(
self, self,
embedding: List[float], embedding: Optional[List[float]],
k: int = 4, k: int = 4,
filter: Optional[Union[dict, list]] = None, filter: Optional[Union[dict, list]] = None,
predicates: Optional[Predicates] = None, predicates: Optional[Predicates] = None,
@ -540,7 +551,7 @@ class TimescaleVector(VectorStore):
async def asimilarity_search_by_vector( async def asimilarity_search_by_vector(
self, self,
embedding: List[float], embedding: Optional[List[float]],
k: int = 4, k: int = 4,
filter: Optional[Union[dict, list]] = None, filter: Optional[Union[dict, list]] = None,
predicates: Optional[Predicates] = None, predicates: Optional[Predicates] = None,