Use Vector DB

To help you create and use LLMs, we are providing a Weaviate database that can be used by the pipelines of the Platform. More informations on the Weaviate database itself on the official site.

Create a VectorDB

To use a Weaviate in your Environment, you need to specify it at the environment creation. There is currently no possibility to add a Weaviate DB once the environment is created. You just need to select the value in the drop-down menu in the Environment Creation form. It will run in a pod near your processing environment and Weaviate will be installed to run its prcessing on the CPU.

Access the VectorDB

Then, you need to parameter some code to access Weaviate's features.

import os
from craft_ai_sdk import CraftAiSdk
from dotenv import load_dotenv

load_dotenv()

sdk = CraftAiSdk(
    sdk_token=CRAFT_AI_SDK_TOKEN,
    environment_url=CRAFT_AI_ENVIRONMENT_URL,
)

client = sdk.get_weaviate_client()

Use the VectorDB

Once you are connected, you can use your VectorDB as you wish and call all its internal functions.

vectordb_collection_name = "your_collection"

weaviate_collection = client.collections.get(
    name=vectordb_collection_name,
)