Hugging Face — Integration

Overview

Create embeddings and call hosted models through the Inference API; AI Hub tries the pipeline endpoint first, then model URL fallback.

Capabilities

• Embeddings via feature-extraction for RAG/search

• Model override and base URL override options

• Batch encode texts via /api/handler/execute task=encode

1. 1

   Step 1

   Save token: POST /AdminApi.php {"provider":"huggingface","creds":{"api_key":"hf_…"}}.

   Back to top
2. 2

   Step 2

   Encode vectors: POST /api/handler/execute {"task":"encode","texts":["hello","world"],"model":"sentence-transformers/all-mpnet-base-v2"}

   Back to top
3. 3

   Step 3

   Persist vectors in your app’s store and attach ids for retrieval.

   Back to top

Limitations

• Embedding dimensionality depends on chosen model; verify before indexing.

• Provider rate limits apply; use the Jobs queue for large batches.

FAQs

Pricing