Cleora returns the computed embeddings in a form of a .npz file.

Embedding Download Interface

The file contains an entity id array under the key entity_id and a corresponding embedding matrix under the key vectors.

>>> cleora_output = np.load('users_oN0fvl26yrcmN6GiN0G32A2kEbH3_LFgcVptT4lV7yq9wv6vf_output_embeddings.npz')

>>>> entity_ids = cleora_output['entity_id']
>>>> entity_ids.shape

>>>> embeddings = cleora_output['vectors']
>>>> embeddings.shape
(171002, 128)

The entity id array corresponds to identifiers from Column 2 in the input file (see Input Data format).

For further advice on the application of the embeddings to various business scenarios, refer to How To Use the Embeddings