Overview
This node interacts with SiliconFlow AI models to perform various AI-related tasks. Specifically, for the Embeddings - Create operation, it generates vector embeddings from input text using a selected embedding model. Embeddings are numerical representations of text that capture semantic meaning, useful for tasks like similarity search, clustering, or as input features for machine learning models.
Common scenarios where this node is beneficial include:
- Converting textual data into embeddings for semantic search engines.
- Preparing text data for downstream AI/ML workflows.
- Generating embeddings for document classification or recommendation systems.
For example, you can input a product description and get its embedding vector to find similar products by comparing vectors.
Properties
| Name | Meaning |
|---|---|
| Model | The embedding model to use. Options include: - BAAI/bge-large-zh-v1.5 (Chinese, 512 tokens) - BAAI/bge-large-en-v1.5 (English, 512 tokens) - BAAI/bge-m3 (Multilingual, 8192 tokens) - Pro/BAAI/bge-m3 (Professional multilingual, 8192 tokens) - Qwen3-Embedding-8B (32768 tokens) - Qwen3-Embedding-4B (32768 tokens) - Qwen3-Embedding-0.6B (32768 tokens) - netease-youdao/bce-embedding-base_v1 (512 tokens) - sentence-transformers/all-MiniLM-L6-v2 |
| Input | The input text string to embed. |
| Encoding Format | The format in which to return the embeddings: - Float (numerical array) - Base64 (encoded string) |
Output
The output JSON contains:
embeddings: An array of embedding vectors corresponding to the input(s). Each embedding is either an array of floats or a base64-encoded string depending on the encoding format chosen.model: The name of the embedding model used.usage: Information about token usage for the request._rawResponse: The full raw response from the SiliconFlow API for debugging or advanced use.
If multiple inputs were processed, embeddings will contain multiple embedding vectors accordingly.
Dependencies
- Requires an active SiliconFlow API key credential configured in n8n.
- Makes HTTP POST requests to the SiliconFlow API endpoint
/embeddings. - Requires network access to the SiliconFlow service base URL specified in credentials.
Troubleshooting
- No response received from the model: This error may occur if the API returns no embedding data. Check your API key validity and ensure the input text is not empty.
- At least one document must be provided: Although related to rerank operation, users mixing operations should ensure required fields are filled.
- Invalid model selection: Ensure the selected embedding model supports the input language and token length.
- API authentication errors: Verify that the API key credential is correctly set up and has necessary permissions.
- Input too long: Some models have token limits (e.g., 512 or 8192 tokens). Inputs exceeding these limits may cause errors.
Links and References
- SiliconFlow API Documentation (hypothetical link)
- Understanding Text Embeddings
- Sentence Transformers Models