TheBobBob commited on
Commit
c012ee6
·
verified ·
1 Parent(s): 037e21e

Delete createVectorDB.py

Browse files
Files changed (1) hide show
  1. createVectorDB.py +0 -49
createVectorDB.py DELETED
@@ -1,49 +0,0 @@
1
- import chromadb
2
- from chromadb.utils import embedding_functions
3
- from typing import Optional, Dict
4
-
5
- def createVectorDB(
6
- collection_name: Optional[str],
7
- chroma_data_path: Optional[str] = None,
8
- embed_model: Optional[str] = "all-MiniLM-L6-v2",
9
- metadata: Optional[Dict[str, str]] = None
10
- ) -> chromadb.Collection:
11
- """Creates the vector database to store embeddings.
12
-
13
- Args:
14
- collection_name (str): The name of the collection.
15
- chroma_data_path (Optional[str]): Path for chroma embeddings.
16
- embed_model (Optional[str]): Model name for embeddings.
17
- metadata (Optional[Dict[str, str]]): Metadata for the collection.
18
-
19
- Returns:
20
- chromadb.Collection: The created collection object.
21
- """
22
- if chroma_data_path is None:
23
- chroma_data_path = r"CHROMA_EMBEDDINGS_PATH" # Default path if not provided
24
-
25
- client = chromadb.PersistentClient(path=chroma_data_path)
26
-
27
- embedding_func = embedding_functions.SentenceTransformerEmbeddingFunction(
28
- model_name=embed_model
29
- )
30
-
31
- # Use provided metadata or default to empty dictionary
32
- if metadata is None:
33
- metadata = {"hnsw:space": "cosine"}
34
-
35
- collection = client.create_collection(
36
- collection_name=collection_name,
37
- embedding_function=embedding_func,
38
- metadata=metadata,
39
- )
40
-
41
- return collection
42
-
43
- #unsure how to create unittest
44
-
45
- #collection = createVectorDB(
46
- #COLLECTION_NAME="123456789",
47
- #C#HROMA_DATA_PATH=r"C:\Users\navan\Downloads\BioModelsRAG\CHROMA_EMBEDDINGS_PATH",
48
- #EMBED_MODEL="all-MiniLM-L6-v2",
49
- #metadata={"hnsw:space": "cosine"}