diff --git a/llama-index-integrations/vector_stores/llama-index-vector-stores-qdrant/llama_index/vector_stores/qdrant/base.py b/llama-index-integrations/vector_stores/llama-index-vector-stores-qdrant/llama_index/vector_stores/qdrant/base.py
index c69ce034e3c68af8de3a5e210f282bc2429fe3bf..151de9d4e18aa71f35f7e9649f0b4ef803e54f39 100644
--- a/llama-index-integrations/vector_stores/llama-index-vector-stores-qdrant/llama_index/vector_stores/qdrant/base.py
+++ b/llama-index-integrations/vector_stores/llama-index-vector-stores-qdrant/llama_index/vector_stores/qdrant/base.py
@@ -29,6 +29,7 @@ from llama_index.vector_stores.qdrant.utils import (
SparseEncoderCallable,
default_sparse_encoder,
relative_score_fusion,
+ fastembed_sparse_encoder,
)
from qdrant_client.http import models as rest
from qdrant_client.http.exceptions import UnexpectedResponse
@@ -49,6 +50,10 @@ import_err_msg = (
"`qdrant-client` package not found, please run `pip install qdrant-client`"
)
+DENSE_VECTOR_NAME = "text-dense"
+SPARSE_VECTOR_NAME_OLD = "text-sparse"
+SPARSE_VECTOR_NAME = "text-sparse-new"
+
class QdrantVectorStore(BasePydanticVectorStore):
"""
@@ -163,11 +168,12 @@ class QdrantVectorStore(BasePydanticVectorStore):
# setup hybrid search if enabled
if enable_hybrid:
- self._sparse_doc_fn = sparse_doc_fn or default_sparse_encoder(
- "naver/efficient-splade-VI-BT-large-doc"
+ self._sparse_doc_fn = sparse_doc_fn or self.get_default_sparse_doc_encoder(
+ collection_name
)
- self._sparse_query_fn = sparse_query_fn or default_sparse_encoder(
- "naver/efficient-splade-VI-BT-large-query"
+ self._sparse_query_fn = (
+ sparse_query_fn
+ or self.get_default_sparse_query_encoder(collection_name)
)
self._hybrid_fusion_fn = hybrid_fusion_fn or cast(
HybridFusionCallable, relative_score_fusion
@@ -228,17 +234,18 @@ class QdrantVectorStore(BasePydanticVectorStore):
):
vectors.append(
{
- "text-sparse": rest.SparseVector(
+ # Dynamically switch between the old and new sparse vector name
+ self.sparse_vector_name: rest.SparseVector(
indices=sparse_indices[i],
values=sparse_vectors[i],
),
- "text-dense": node.get_embedding(),
+ DENSE_VECTOR_NAME: node.get_embedding(),
}
)
else:
vectors.append(
{
- "text-dense": node.get_embedding(),
+ DENSE_VECTOR_NAME: node.get_embedding(),
}
)
else:
@@ -375,13 +382,14 @@ class QdrantVectorStore(BasePydanticVectorStore):
self._client.create_collection(
collection_name=collection_name,
vectors_config={
- "text-dense": rest.VectorParams(
+ DENSE_VECTOR_NAME: rest.VectorParams(
size=vector_size,
distance=rest.Distance.COSINE,
)
},
+ # Newly created collection will have the new sparse vector name
sparse_vectors_config={
- "text-sparse": rest.SparseVectorParams(
+ SPARSE_VECTOR_NAME: rest.SparseVectorParams(
index=rest.SparseIndexParams()
)
},
@@ -413,13 +421,13 @@ class QdrantVectorStore(BasePydanticVectorStore):
await self._aclient.create_collection(
collection_name=collection_name,
vectors_config={
- "text-dense": rest.VectorParams(
+ DENSE_VECTOR_NAME: rest.VectorParams(
size=vector_size,
distance=rest.Distance.COSINE,
)
},
sparse_vectors_config={
- "text-sparse": rest.SparseVectorParams(
+ SPARSE_VECTOR_NAME: rest.SparseVectorParams(
index=rest.SparseIndexParams()
)
},
@@ -497,7 +505,7 @@ class QdrantVectorStore(BasePydanticVectorStore):
requests=[
rest.SearchRequest(
vector=rest.NamedVector(
- name="text-dense",
+ name=DENSE_VECTOR_NAME,
vector=query_embedding,
),
limit=query.similarity_top_k,
@@ -506,7 +514,8 @@ class QdrantVectorStore(BasePydanticVectorStore):
),
rest.SearchRequest(
vector=rest.NamedSparseVector(
- name="text-sparse",
+ # Dynamically switch between the old and new sparse vector name
+ name=self.sparse_vector_name,
vector=rest.SparseVector(
indices=sparse_indices[0],
values=sparse_embedding[0],
@@ -548,7 +557,8 @@ class QdrantVectorStore(BasePydanticVectorStore):
requests=[
rest.SearchRequest(
vector=rest.NamedSparseVector(
- name="text-sparse",
+ # Dynamically switch between the old and new sparse vector name
+ name=self.sparse_vector_name,
vector=rest.SparseVector(
indices=sparse_indices[0],
values=sparse_embedding[0],
@@ -569,7 +579,7 @@ class QdrantVectorStore(BasePydanticVectorStore):
requests=[
rest.SearchRequest(
vector=rest.NamedVector(
- name="text-dense",
+ name=DENSE_VECTOR_NAME,
vector=query_embedding,
),
limit=query.similarity_top_k,
@@ -629,7 +639,7 @@ class QdrantVectorStore(BasePydanticVectorStore):
requests=[
rest.SearchRequest(
vector=rest.NamedVector(
- name="text-dense",
+ name=DENSE_VECTOR_NAME,
vector=query_embedding,
),
limit=query.similarity_top_k,
@@ -638,7 +648,8 @@ class QdrantVectorStore(BasePydanticVectorStore):
),
rest.SearchRequest(
vector=rest.NamedSparseVector(
- name="text-sparse",
+ # Dynamically switch between the old and new sparse vector name
+ name=self.sparse_vector_name,
vector=rest.SparseVector(
indices=sparse_indices[0],
values=sparse_embedding[0],
@@ -679,7 +690,8 @@ class QdrantVectorStore(BasePydanticVectorStore):
requests=[
rest.SearchRequest(
vector=rest.NamedSparseVector(
- name="text-sparse",
+ # Dynamically switch between the old and new sparse vector name
+ name=self.sparse_vector_name,
vector=rest.SparseVector(
indices=sparse_indices[0],
values=sparse_embedding[0],
@@ -699,7 +711,7 @@ class QdrantVectorStore(BasePydanticVectorStore):
requests=[
rest.SearchRequest(
vector=rest.NamedVector(
- name="text-dense",
+ name=DENSE_VECTOR_NAME,
vector=query_embedding,
),
limit=query.similarity_top_k,
@@ -857,3 +869,34 @@ class QdrantVectorStore(BasePydanticVectorStore):
)
return Filter(must=must_conditions)
+
+ def use_old_sparse_encoder(self, collection_name: str) -> bool:
+ return (
+ self._collection_exists(collection_name)
+ and SPARSE_VECTOR_NAME_OLD
+ in self.client.get_collection(collection_name).config.params.vectors
+ )
+
+ @property
+ def sparse_vector_name(self) -> str:
+ return (
+ SPARSE_VECTOR_NAME_OLD
+ if self.use_old_sparse_encoder(self.collection_name)
+ else SPARSE_VECTOR_NAME
+ )
+
+ def get_default_sparse_doc_encoder(
+ self, collection_name: str
+ ) -> SparseEncoderCallable:
+ if self.use_old_sparse_encoder(collection_name):
+ return default_sparse_encoder("naver/efficient-splade-VI-BT-large-doc")
+
+ return fastembed_sparse_encoder(model_name="prithvida/Splade_PP_en_v1")
+
+ def get_default_sparse_query_encoder(
+ self, collection_name: str
+ ) -> SparseEncoderCallable:
+ if self.use_old_sparse_encoder(collection_name):
+ return default_sparse_encoder("naver/efficient-splade-VI-BT-large-query")
+
+ return fastembed_sparse_encoder(model_name="prithvida/Splade_PP_en_v1")
diff --git a/llama-index-integrations/vector_stores/llama-index-vector-stores-qdrant/pyproject.toml b/llama-index-integrations/vector_stores/llama-index-vector-stores-qdrant/pyproject.toml
index 08d37b7ad52d842409626f28023a88537d95554d..1a0f0c9b4cab49cc02f87d65c8ee00847225c661 100644
--- a/llama-index-integrations/vector_stores/llama-index-vector-stores-qdrant/pyproject.toml
+++ b/llama-index-integrations/vector_stores/llama-index-vector-stores-qdrant/pyproject.toml
@@ -27,7 +27,7 @@ exclude = ["**/BUILD"]
license = "MIT"
name = "llama-index-vector-stores-qdrant"
readme = "README.md"
-version = "0.1.6"
+version = "0.2.0"
[tool.poetry.dependencies]
python = ">=3.9,<3.13"