Blab Embed for Agent
Embedding Model for Vector DB - Upstage Solar Embeddings. Supports up to 100 strings per request with max 204,800 total tokens. Each text should be under 4000 tokens (optimal: under 512 tokens).
Overview
This node generates text embeddings using the Upstage Solar Embeddings API. It supports embedding up to 100 strings per request, with each string ideally under 512 tokens. The node is useful for transforming text data into vector representations for use in AI vector stores, search queries, or document analysis. Practical applications include semantic search, question answering, and document similarity analysis.
Use Case Examples
- Embedding search queries to improve search relevance.
- Embedding passages or documents for vector-based document retrieval.
Properties
| Name | Meaning |
|---|---|
| Model | The Upstage embedding model to use, optimized either for search queries and questions or for documents and passages. |
Output
JSON
response- The array of generated embeddings corresponding to the input texts.
Dependencies
- Requires an API key credential for Upstage API authentication.
Troubleshooting
- Error 'No valid input texts provided for embedding' indicates that the input text array is empty or contains only invalid strings; ensure input texts are non-empty strings.
- Error 'Too many texts' means the input exceeds the maximum of 100 strings per request; reduce the batch size accordingly.
- API errors with status codes indicate issues with the API request or credentials; verify the API key and request format.
- Mismatch in expected embeddings count suggests a problem with the API response; retry or check API status.
Links
- LangChain Text Embedding Documentation - Primary documentation for text embedding models used in this node.