Upstash Vector Stores

​

Node Details

• Name: Upstash_VectorStores
• Type: Upstash
• Version: 2.0
• Category: Vector Stores

​

Base Classes

• Upstash
• VectorStoreRetriever
• BaseRetriever

​

Credentials

• Type: upstashVectorApi
• Required Fields:
  • UPSTASH_VECTOR_REST_URL
  • UPSTASH_VECTOR_REST_TOKEN

​

Input Parameters

1. Document (optional, list)
   • Type: Document
   • Description: List of documents to be processed
2. Embeddings (required)
   • Type: Embeddings
   • Description: Embedding model to use
3. Record Manager (optional)
   • Type: RecordManager
   • Description: Keeps track of records to prevent duplication
4. File Upload (optional)
   • Type: boolean
   • Description: Allows file upload on the chat
5. Upstash Metadata Filter (optional)
   • Type: string
   • Description: Filter for Upstash metadata
6. Top K (optional)
   • Type: number
   • Description: Number of top results to fetch (default: 4)

​

Outputs

1. Upstash Retriever
   • Type: retriever
   • Base Classes: [Upstash, VectorStoreRetriever, BaseRetriever]
2. Upstash Vector Store
   • Type: vectorStore
   • Base Classes: [Upstash, UpstashVectorStore base classes]

​

Functionality

1. Upsert:
   • Processes input documents
   • Creates or updates embeddings in the Upstash vector store
   • Handles file uploads and chat IDs if enabled
   • Uses RecordManager for deduplication if provided
2. Delete:
   • Removes specified documents from the vector store
   • Supports deletion using RecordManager or direct ID-based deletion
3. Initialization:
   • Sets up the Upstash vector store with provided credentials
   • Applies metadata filters and chat ID filters if specified
   • Creates a retriever or vector store based on the configuration

​

Use Cases

• Storing and retrieving document embeddings for similarity search
• Managing document collections with metadata filtering
• Integrating file upload capabilities in chat applications
• Implementing efficient document retrieval systems

​

Notes

• The node supports both standalone usage and integration with a RecordManager for more advanced document tracking.
• File upload feature can be enabled to associate documents with specific chat sessions.
• The node uses the UpstashVectorStore from the LangChain library for vector operations.