Add embeding_dims param to FAISS (#2513)

Author: Dev-Khant

mem0/configs/vector_stores/faiss.py

@@ -12,6 +12,7 @@ class FAISSConfig(BaseModel):
     normalize_L2: bool = Field(
         False, description="Whether to normalize L2 vectors (only applicable for euclidean distance)"
     )
+    embedding_model_dims: int = Field(1536, description="Dimension of the embedding vector")
 
     @model_validator(mode="before")
     @classmethod

mem0/vector_stores/faiss.py

@@ -35,6 +35,7 @@ def __init__(
         path: Optional[str] = None,
         distance_strategy: str = "euclidean",
         normalize_L2: bool = False,
+        embedding_model_dims: int = 1536,
     ):
         """
         Initialize the FAISS vector store.
@@ -51,6 +52,7 @@ def __init__(
         self.path = path or f"/tmp/faiss/{collection_name}"
         self.distance_strategy = distance_strategy
         self.normalize_L2 = normalize_L2
+        self.embedding_model_dims = embedding_model_dims
 
         # Initialize storage structures
         self.index = None
@@ -145,13 +147,12 @@ def _parse_output(self, scores, ids, limit=None) -> List[OutputData]:
 
         return results
 
-    def create_col(self, name: str, vector_size: int = 1536, distance: str = None):
+    def create_col(self, name: str, distance: str = None):
         """
         Create a new collection.
 
         Args:
             name (str): Name of the collection.
-            vector_size (int, optional): Dimensionality of vectors. Defaults to 1536.
             distance (str, optional): Distance metric to use. Overrides the distance_strategy
                 passed during initialization. Defaults to None.
 
@@ -162,9 +163,9 @@ def create_col(self, name: str, vector_size: int = 1536, distance: str = None):
 
         # Create index based on distance strategy
         if distance_strategy.lower() == "inner_product" or distance_strategy.lower() == "cosine":
-            self.index = faiss.IndexFlatIP(vector_size)
+            self.index = faiss.IndexFlatIP(self.embedding_model_dims)
         else:
-            self.index = faiss.IndexFlatL2(vector_size)
+            self.index = faiss.IndexFlatL2(self.embedding_model_dims)
 
         self.collection_name = name
 

pyproject.toml

@@ -1,6 +1,6 @@
 [tool.poetry]
 name = "mem0ai"
-version = "0.1.85"
+version = "0.1.86"
 description = "Long-term memory for AI Agents"
 authors = ["Mem0 <[email protected]>"]
 exclude = [