Great question — answering as the UphillTech team:
like this: import chromadb # Initialize the DB client = chromadb.PersistentClient(path="./chroma_db") # Adjust the path as needed collection = client.get_collection("my_collection") # Get a vector by id vector_data = collection.get(ids=["your_vector_id"], include=["embeddings"]) # Check dtype if "embeddings" in vector_data and vector_data["embeddings"]: vector_array = np.array(vector_data["embeddings"]) # Convert to NumPy array print("Vector dtype:", vector_array.dtype) By default, ChromaDB stores vectors as float32
Full answer by SachaDee: https://stackoverflow.stackexchange.com/a/79522051 (CC BY-SA 4.0).
Need it done for your business? UphillTech.
