EDB Embeddings v1.4.0 (LTS)

The EDB Embeddings component connects a flow to an embedding model running on a Hybrid Manager (HM) model server cluster, and produces an Embeddings model object that downstream components (vector stores, retrievers) can use to encode text.

Choosing the right component

Use the EDB Embeddings component when:

  • You want vector embeddings produced by a model running on an HM-hosted model server cluster (OpenAI-compatible or NVIDIA NIM).
  • You're feeding a vector store, retriever, or similarity-search component that expects an Embeddings model.

Use a different component if:

  • You want CPU-based embeddings produced locally on the database cluster using aidb. Use EDB Embedded Models.
  • You want generative responses from a model, not embeddings. Use EDB Model Server.

Prerequisites

  • An HM model server cluster with an embedding model deployed. See Models for how to create one. A GPU node is typically required.

  • An HM machine-user access key saved in Langflow as a Global Variable (default name HM_API_KEY).

Inputs

Connection

FieldTypeRequiredDefaultNotes
Hybrid Manager URLTextNoAdvanced. Override the default HM URL.
HM Machine User KeySecretYesHM_API_KEYDefaults to the global variable named HM_API_KEY.
Hybrid Manager Model Server Cluster InstanceDropdownYesThe model server cluster to call. Populated from your HM model clusters.
External IngressBooleanNoAdvanced. Route through the external ingress instead of the in-cluster service.

Model

FieldTypeRequiredDefaultNotes
API ClientDropdownNoOpenAIAdvanced. OpenAI for the EDB Model Server's OpenAI-compatible API. NVIDIA only when connecting to a native NIM endpoint that needs NVIDIA-specific embedding features.
ModelDropdownYesPopulated from the selected model server cluster. Use the refresh button if the list is empty.

API Client = OpenAI (default)

FieldTypeRequiredDefaultNotes
DimensionsIntegerNoAdvanced. Number of dimensions for the output vectors. Only supported by certain models. Leave blank to use the model's default.
Max RetriesIntegerNo3Advanced. Maximum retries on a failed embedding call.
Chunk SizeIntegerNo1000Advanced. Maximum texts to send in a single batch.
Model KwargsDictNoAdvanced. Additional keyword arguments passed through to the OpenAI embeddings client.

API Client = NVIDIA

FieldTypeRequiredDefaultNotes
Model TemperatureFloatNo0.1Advanced. Model temperature.

The OpenAI-specific fields hide when API Client is NVIDIA, and vice versa. Field values are preserved when you switch between clients, so a round-trip doesn't wipe your settings.

Outputs

OutputTypeCarries
Embedding ModelEmbeddingsA LangChain Embeddings instance pointing at the selected HM model server cluster. Pass this into a vector store or retriever.