HuggingFace Embeddings (Local)
Configure local embedding models with HuggingFace
Configuration
embedding_model:
provider: "huggingface_local"
model: "sentence-transformers/all-mpnet-base-v2"
model_kwargs: {}
encode_kwargs: {}Configuration Options
| Field | Type | Default | Required | Description |
|---|---|---|---|---|
provider | string | - | ✅ | Must be "huggingface_local" |
model | string | "sentence-transformers/all-mpnet-base-v2" | ❌ | Model path or name |
model_kwargs | object | ❌ | Model initialization parameters | |
encode_kwargs | object | ❌ | Encoding parameters | |
rpm | integer | 3000 | ❌ | Requests per minute |
tpm | integer | 1000000 | ❌ | Tokens per minute |
Setup
- Local models download automatically on first use
- GPU support: Models use available CUDA/MPS if present
- Offline operation: No internet required after download