Expand description
Request and response model types for the embedding API endpoints.
Dense types are OpenAI-compatible. Sparse and dual types are BGE-M3 specific; they extend the same request shape with additional output fields.
Structsยง
- Dense
Embedding Data - Per-document dense embedding entry.
- Dense
Request - Request body for the dense embeddings endpoint.
- Dense
Response - Top-level response for the dense embeddings endpoint (OpenAI-compatible).
- Dual
Embedding Data - Per-document paired dense + sparse embedding entry.
- Dual
Request - Request body for the unified dense + sparse embeddings endpoint.
- Dual
Response - Top-level response for the unified dense + sparse embeddings endpoint.
- Model
Entry - A single model entry.
- Models
Response - Top-level response for GET /v1/models (OpenAI-compatible).
- Sparse
Embedding Data - Per-document sparse embedding entry.
- Sparse
Request - Request body for the sparse embeddings endpoint.
- Sparse
Response - Top-level response for the sparse embeddings endpoint.
- Sparse
Values - Parallel arrays of token indices and their weights.
- Text
Input - Newtype wrapping a
Vec<String>that deserializes from either"a single string"or["array", "of", "strings"]. - Usage
- Token usage counters.