Vector Store

What is a Vector Store?

A Vector Store is a specialized database designed to store and manage vector embeddings of text data. Vector Store acts as your central hub for all LLM-related vector embeddings within LLM Labs. It offers:

• Efficient Retrieval: It allows you to quickly retrieve relevant information based on similarity score, making it ideal for tasks like question answering and document search.

• Knowledge Base: The Vector Store can be enhanced with additional information, turning vectors into rich knowledge representations.

• Flexibility: You can easily create, update, and delete vectors, providing flexibility for various projects.

Getting Started

You can visit the Vector stores page by selecting the Vector stores option located in the LLM Labs sidebar.

Vector Store Creation

1. Click the Create new vector store button.

2. Configure your vector store settings.

The configurations are:

• Name: Your preferred vector store name

• Vector store provider: We support two types of vector store provider:

  • Datasaur: Utilizing the Datasaur provider means Datasaur will handle the embedding process for you.

  • External: Choosing the External provider allows you to handle the embedding process independently.

• Embedding model: Your preferred embedding models. Several embedding models that we support by default are:

  • text-embedding-ada-002

  • text-embedding-3-small

  • text-embedding-3-large

• Chunk size: The maximum number of characters that a chunk can contain. The larger the numbers, the bigger each chunk will be, allowing more data to be included within it.

• Overlap: The number of characters that should overlap between two adjacent chunks. The larger the overlap, the more information each chunk shares with its neighboring chunks.

• Advanced settings: Additional settings can enhance your data organization by enabling you to provide information about the file using the File Properties feature.

Knowledge Base

1. Once the vector store is created, you can add your files for embedding by uploading them to the Knowledge Base. You can also add your files via the External Object Storage.

2. After you select the files, please click on the "Update Vector Store" button to initiate the embedding process. The embedding process might take some time, depending on the file size and the number of files.

3. After completing the embedding process, you can preview the files and use them to conduct Retrieval-Augmented Generation (RAG) in LLM Labs. In this example, we embed our sample "Patient Records," which will be used for the RAG process in LLM Labs.

Search

The search function allows you to validate the effectiveness of your knowledge base in providing context. The search results are shown in chunks that follow the chunk size and overlap value you specified when creating the vector store. Each chunk will have a similarity score along with its source. A higher similarity score means the chunk content is more related to the given prompt.

Activity

The Activity feature logs all actions performed on your Vector stores, making it easier to track changes and actions. You can filter the activity based on member, file, file source, and date.

RAG Example: Healthcare Assistant

Here is how Vector Store can streamline the development of a Retrieval-Augmented Generation (RAG) based Healthcare Assistant in LLM Labs:

1. Create the LLM Playground with the User Instruction and System Instruction you've prepared.

2. From the Vector stores dropdown, select the vector store you created.

3. Write your prompt asking about a patient's health condition. The results from the vector store will then be displayed.

4. You can also view the corresponding chunks from the vector store and the source.

This is just one example! Vector Store empowers you to build various LLM applications that rely on efficient retrieval of semantically related information.

Ready to Streamline Your Workflow?

Explore the LLM Labs documentation for detailed instructions based on your plan and functionalities. Contact us at [e-mail] if you need further assistance, our support team is always happy to help!