What are embedding outputs?

Embedding outputs are learned vector representations of entities in the graph. Entities with similar learned behavior or context tend to have similar representations.

You can use these vectors in downstream workflows such as clustering, segmentation, retrieval, nearest-neighbor search, or as inputs to another model. For example, customer embeddings can support cohort discovery, while product embeddings can support similarity search.

Select embeddings as an output type when configuring the batch prediction, then choose a supported output destination. You can request predictions, embeddings, or both.

See Batch Predictions and Prediction Outputs.