yangdx
commited on
Commit
·
8aa0a5e
1
Parent(s):
c6b21a9
refactor: make cosine similarity threshold a required config parameter
Browse files• Remove default threshold from env var
• Add validation for missing threshold
• Move default to lightrag.py config init
• Update all vector DB implementations
• Improve threshold validation consistency
- lightrag/kg/chroma_impl.py +5 -5
- lightrag/kg/faiss_impl.py +5 -4
- lightrag/kg/milvus_impl.py +9 -1
- lightrag/kg/nano_vector_db_impl.py +5 -4
- lightrag/kg/oracle_impl.py +5 -5
- lightrag/kg/postgres_impl.py +5 -5
- lightrag/kg/qdrant_impl.py +11 -1
- lightrag/kg/tidb_impl.py +5 -5
- lightrag/lightrag.py +9 -0
lightrag/kg/chroma_impl.py
CHANGED
@@ -13,15 +13,15 @@ from lightrag.utils import logger
|
|
13 |
class ChromaVectorDBStorage(BaseVectorStorage):
|
14 |
"""ChromaDB vector storage implementation."""
|
15 |
|
16 |
-
cosine_better_than_threshold: float =
|
17 |
|
18 |
def __post_init__(self):
|
19 |
try:
|
20 |
-
# Use global config value if specified, otherwise use default
|
21 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
22 |
-
|
23 |
-
|
24 |
-
|
|
|
25 |
|
26 |
user_collection_settings = config.get("collection_settings", {})
|
27 |
# Default HNSW index settings for ChromaDB
|
|
|
13 |
class ChromaVectorDBStorage(BaseVectorStorage):
|
14 |
"""ChromaDB vector storage implementation."""
|
15 |
|
16 |
+
cosine_better_than_threshold: float = None
|
17 |
|
18 |
def __post_init__(self):
|
19 |
try:
|
|
|
20 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
21 |
+
cosine_threshold = config.get("cosine_better_than_threshold")
|
22 |
+
if cosine_threshold is None:
|
23 |
+
raise ValueError("cosine_better_than_threshold must be specified in vector_db_storage_cls_kwargs")
|
24 |
+
self.cosine_better_than_threshold = cosine_threshold
|
25 |
|
26 |
user_collection_settings = config.get("collection_settings", {})
|
27 |
# Default HNSW index settings for ChromaDB
|
lightrag/kg/faiss_impl.py
CHANGED
@@ -23,14 +23,15 @@ class FaissVectorDBStorage(BaseVectorStorage):
|
|
23 |
Uses cosine similarity by storing normalized vectors in a Faiss index with inner product search.
|
24 |
"""
|
25 |
|
26 |
-
cosine_better_than_threshold: float =
|
27 |
|
28 |
def __post_init__(self):
|
29 |
# Grab config values if available
|
30 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
31 |
-
|
32 |
-
|
33 |
-
|
|
|
34 |
|
35 |
# Where to save index file if you want persistent storage
|
36 |
self._faiss_index_file = os.path.join(
|
|
|
23 |
Uses cosine similarity by storing normalized vectors in a Faiss index with inner product search.
|
24 |
"""
|
25 |
|
26 |
+
cosine_better_than_threshold: float = None
|
27 |
|
28 |
def __post_init__(self):
|
29 |
# Grab config values if available
|
30 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
31 |
+
cosine_threshold = config.get("cosine_better_than_threshold")
|
32 |
+
if cosine_threshold is None:
|
33 |
+
raise ValueError("cosine_better_than_threshold must be specified in vector_db_storage_cls_kwargs")
|
34 |
+
self.cosine_better_than_threshold = cosine_threshold
|
35 |
|
36 |
# Where to save index file if you want persistent storage
|
37 |
self._faiss_index_file = os.path.join(
|
lightrag/kg/milvus_impl.py
CHANGED
@@ -19,6 +19,8 @@ config.read("config.ini", "utf-8")
|
|
19 |
|
20 |
@dataclass
|
21 |
class MilvusVectorDBStorge(BaseVectorStorage):
|
|
|
|
|
22 |
@staticmethod
|
23 |
def create_collection_if_not_exist(
|
24 |
client: MilvusClient, collection_name: str, **kwargs
|
@@ -30,6 +32,12 @@ class MilvusVectorDBStorge(BaseVectorStorage):
|
|
30 |
)
|
31 |
|
32 |
def __post_init__(self):
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
self._client = MilvusClient(
|
34 |
uri=os.environ.get(
|
35 |
"MILVUS_URI",
|
@@ -103,7 +111,7 @@ class MilvusVectorDBStorge(BaseVectorStorage):
|
|
103 |
data=embedding,
|
104 |
limit=top_k,
|
105 |
output_fields=list(self.meta_fields),
|
106 |
-
search_params={"metric_type": "COSINE", "params": {"radius":
|
107 |
)
|
108 |
print(results)
|
109 |
return [
|
|
|
19 |
|
20 |
@dataclass
|
21 |
class MilvusVectorDBStorge(BaseVectorStorage):
|
22 |
+
cosine_better_than_threshold: float = None
|
23 |
+
|
24 |
@staticmethod
|
25 |
def create_collection_if_not_exist(
|
26 |
client: MilvusClient, collection_name: str, **kwargs
|
|
|
32 |
)
|
33 |
|
34 |
def __post_init__(self):
|
35 |
+
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
36 |
+
cosine_threshold = config.get("cosine_better_than_threshold")
|
37 |
+
if cosine_threshold is None:
|
38 |
+
raise ValueError("cosine_better_than_threshold must be specified in vector_db_storage_cls_kwargs")
|
39 |
+
self.cosine_better_than_threshold = cosine_threshold
|
40 |
+
|
41 |
self._client = MilvusClient(
|
42 |
uri=os.environ.get(
|
43 |
"MILVUS_URI",
|
|
|
111 |
data=embedding,
|
112 |
limit=top_k,
|
113 |
output_fields=list(self.meta_fields),
|
114 |
+
search_params={"metric_type": "COSINE", "params": {"radius": self.cosine_better_than_threshold}},
|
115 |
)
|
116 |
print(results)
|
117 |
return [
|
lightrag/kg/nano_vector_db_impl.py
CHANGED
@@ -73,16 +73,17 @@ from lightrag.base import (
|
|
73 |
|
74 |
@dataclass
|
75 |
class NanoVectorDBStorage(BaseVectorStorage):
|
76 |
-
cosine_better_than_threshold: float =
|
77 |
|
78 |
def __post_init__(self):
|
79 |
# Initialize lock only for file operations
|
80 |
self._save_lock = asyncio.Lock()
|
81 |
# Use global config value if specified, otherwise use default
|
82 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
83 |
-
|
84 |
-
|
85 |
-
|
|
|
86 |
|
87 |
self._client_file_name = os.path.join(
|
88 |
self.global_config["working_dir"], f"vdb_{self.namespace}.json"
|
|
|
73 |
|
74 |
@dataclass
|
75 |
class NanoVectorDBStorage(BaseVectorStorage):
|
76 |
+
cosine_better_than_threshold: float = None
|
77 |
|
78 |
def __post_init__(self):
|
79 |
# Initialize lock only for file operations
|
80 |
self._save_lock = asyncio.Lock()
|
81 |
# Use global config value if specified, otherwise use default
|
82 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
83 |
+
cosine_threshold = config.get("cosine_better_than_threshold")
|
84 |
+
if cosine_threshold is None:
|
85 |
+
raise ValueError("cosine_better_than_threshold must be specified in vector_db_storage_cls_kwargs")
|
86 |
+
self.cosine_better_than_threshold = cosine_threshold
|
87 |
|
88 |
self._client_file_name = os.path.join(
|
89 |
self.global_config["working_dir"], f"vdb_{self.namespace}.json"
|
lightrag/kg/oracle_impl.py
CHANGED
@@ -320,14 +320,14 @@ class OracleKVStorage(BaseKVStorage):
|
|
320 |
class OracleVectorDBStorage(BaseVectorStorage):
|
321 |
# db instance must be injected before use
|
322 |
# db: OracleDB
|
323 |
-
cosine_better_than_threshold: float =
|
324 |
|
325 |
def __post_init__(self):
|
326 |
-
# Use global config value if specified, otherwise use default
|
327 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
328 |
-
|
329 |
-
|
330 |
-
|
|
|
331 |
|
332 |
async def upsert(self, data: dict[str, dict]):
|
333 |
"""向向量数据库中插入数据"""
|
|
|
320 |
class OracleVectorDBStorage(BaseVectorStorage):
|
321 |
# db instance must be injected before use
|
322 |
# db: OracleDB
|
323 |
+
cosine_better_than_threshold: float = None
|
324 |
|
325 |
def __post_init__(self):
|
|
|
326 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
327 |
+
cosine_threshold = config.get("cosine_better_than_threshold")
|
328 |
+
if cosine_threshold is None:
|
329 |
+
raise ValueError("cosine_better_than_threshold must be specified in vector_db_storage_cls_kwargs")
|
330 |
+
self.cosine_better_than_threshold = cosine_threshold
|
331 |
|
332 |
async def upsert(self, data: dict[str, dict]):
|
333 |
"""向向量数据库中插入数据"""
|
lightrag/kg/postgres_impl.py
CHANGED
@@ -299,15 +299,15 @@ class PGKVStorage(BaseKVStorage):
|
|
299 |
class PGVectorStorage(BaseVectorStorage):
|
300 |
# db instance must be injected before use
|
301 |
# db: PostgreSQLDB
|
302 |
-
cosine_better_than_threshold: float =
|
303 |
|
304 |
def __post_init__(self):
|
305 |
self._max_batch_size = self.global_config["embedding_batch_num"]
|
306 |
-
# Use global config value if specified, otherwise use default
|
307 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
308 |
-
|
309 |
-
|
310 |
-
|
|
|
311 |
|
312 |
def _upsert_chunks(self, item: dict):
|
313 |
try:
|
|
|
299 |
class PGVectorStorage(BaseVectorStorage):
|
300 |
# db instance must be injected before use
|
301 |
# db: PostgreSQLDB
|
302 |
+
cosine_better_than_threshold: float = None
|
303 |
|
304 |
def __post_init__(self):
|
305 |
self._max_batch_size = self.global_config["embedding_batch_num"]
|
|
|
306 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
307 |
+
cosine_threshold = config.get("cosine_better_than_threshold")
|
308 |
+
if cosine_threshold is None:
|
309 |
+
raise ValueError("cosine_better_than_threshold must be specified in vector_db_storage_cls_kwargs")
|
310 |
+
self.cosine_better_than_threshold = cosine_threshold
|
311 |
|
312 |
def _upsert_chunks(self, item: dict):
|
313 |
try:
|
lightrag/kg/qdrant_impl.py
CHANGED
@@ -50,6 +50,8 @@ def compute_mdhash_id_for_qdrant(
|
|
50 |
|
51 |
@dataclass
|
52 |
class QdrantVectorDBStorage(BaseVectorStorage):
|
|
|
|
|
53 |
@staticmethod
|
54 |
def create_collection_if_not_exist(
|
55 |
client: QdrantClient, collection_name: str, **kwargs
|
@@ -59,6 +61,12 @@ class QdrantVectorDBStorage(BaseVectorStorage):
|
|
59 |
client.create_collection(collection_name, **kwargs)
|
60 |
|
61 |
def __post_init__(self):
|
|
|
|
|
|
|
|
|
|
|
|
|
62 |
self._client = QdrantClient(
|
63 |
url=os.environ.get(
|
64 |
"QDRANT_URL", config.get("qdrant", "uri", fallback=None)
|
@@ -131,4 +139,6 @@ class QdrantVectorDBStorage(BaseVectorStorage):
|
|
131 |
with_payload=True,
|
132 |
)
|
133 |
logger.debug(f"query result: {results}")
|
134 |
-
|
|
|
|
|
|
50 |
|
51 |
@dataclass
|
52 |
class QdrantVectorDBStorage(BaseVectorStorage):
|
53 |
+
cosine_better_than_threshold: float = None
|
54 |
+
|
55 |
@staticmethod
|
56 |
def create_collection_if_not_exist(
|
57 |
client: QdrantClient, collection_name: str, **kwargs
|
|
|
61 |
client.create_collection(collection_name, **kwargs)
|
62 |
|
63 |
def __post_init__(self):
|
64 |
+
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
65 |
+
cosine_threshold = config.get("cosine_better_than_threshold")
|
66 |
+
if cosine_threshold is None:
|
67 |
+
raise ValueError("cosine_better_than_threshold must be specified in vector_db_storage_cls_kwargs")
|
68 |
+
self.cosine_better_than_threshold = cosine_threshold
|
69 |
+
|
70 |
self._client = QdrantClient(
|
71 |
url=os.environ.get(
|
72 |
"QDRANT_URL", config.get("qdrant", "uri", fallback=None)
|
|
|
139 |
with_payload=True,
|
140 |
)
|
141 |
logger.debug(f"query result: {results}")
|
142 |
+
# 添加余弦相似度过滤
|
143 |
+
filtered_results = [dp for dp in results if dp.score >= self.cosine_better_than_threshold]
|
144 |
+
return [{**dp.payload, "id": dp.id, "distance": dp.score} for dp in filtered_results]
|
lightrag/kg/tidb_impl.py
CHANGED
@@ -212,18 +212,18 @@ class TiDBKVStorage(BaseKVStorage):
|
|
212 |
class TiDBVectorDBStorage(BaseVectorStorage):
|
213 |
# db instance must be injected before use
|
214 |
# db: TiDB
|
215 |
-
cosine_better_than_threshold: float =
|
216 |
|
217 |
def __post_init__(self):
|
218 |
self._client_file_name = os.path.join(
|
219 |
self.global_config["working_dir"], f"vdb_{self.namespace}.json"
|
220 |
)
|
221 |
self._max_batch_size = self.global_config["embedding_batch_num"]
|
222 |
-
# Use global config value if specified, otherwise use default
|
223 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
224 |
-
|
225 |
-
|
226 |
-
|
|
|
227 |
|
228 |
async def query(self, query: str, top_k: int) -> list[dict]:
|
229 |
"""Search from tidb vector"""
|
|
|
212 |
class TiDBVectorDBStorage(BaseVectorStorage):
|
213 |
# db instance must be injected before use
|
214 |
# db: TiDB
|
215 |
+
cosine_better_than_threshold: float = None
|
216 |
|
217 |
def __post_init__(self):
|
218 |
self._client_file_name = os.path.join(
|
219 |
self.global_config["working_dir"], f"vdb_{self.namespace}.json"
|
220 |
)
|
221 |
self._max_batch_size = self.global_config["embedding_batch_num"]
|
|
|
222 |
config = self.global_config.get("vector_db_storage_cls_kwargs", {})
|
223 |
+
cosine_threshold = config.get("cosine_better_than_threshold")
|
224 |
+
if cosine_threshold is None:
|
225 |
+
raise ValueError("cosine_better_than_threshold must be specified in vector_db_storage_cls_kwargs")
|
226 |
+
self.cosine_better_than_threshold = cosine_threshold
|
227 |
|
228 |
async def query(self, query: str, top_k: int) -> list[dict]:
|
229 |
"""Search from tidb vector"""
|
lightrag/lightrag.py
CHANGED
@@ -420,6 +420,15 @@ class LightRAG:
|
|
420 |
# Check environment variables
|
421 |
self.check_storage_env_vars(storage_name)
|
422 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
423 |
# show config
|
424 |
global_config = asdict(self)
|
425 |
_print_config = ",\n ".join([f"{k} = {v}" for k, v in global_config.items()])
|
|
|
420 |
# Check environment variables
|
421 |
self.check_storage_env_vars(storage_name)
|
422 |
|
423 |
+
# Ensure vector_db_storage_cls_kwargs has required fields
|
424 |
+
default_vector_db_kwargs = {
|
425 |
+
"cosine_better_than_threshold": float(os.getenv("COSINE_THRESHOLD", "0.2"))
|
426 |
+
}
|
427 |
+
self.vector_db_storage_cls_kwargs = {
|
428 |
+
**default_vector_db_kwargs,
|
429 |
+
**self.vector_db_storage_cls_kwargs
|
430 |
+
}
|
431 |
+
|
432 |
# show config
|
433 |
global_config = asdict(self)
|
434 |
_print_config = ",\n ".join([f"{k} = {v}" for k, v in global_config.items()])
|